Paperless conference system based on big data

Abstract

The invention discloses a paperless conference system based on big data, which relates to the technical field of conference systems, focuses on considering connection distance and sub-task similarity to screen optimal edge computing nodes, comprehensively evaluates network quality indexes such as physical distance, bandwidth, delay and the like between participants and the edge computing nodes, and improves the conference performance. In addition, a task similarity calculation resource scheduling result based on historical task data is introduced, and it is ensured that the current task can be matched with the most suitable edge calculation node used by the historical task. And finally, according to the state information of each edge computing node and the reference demand characteristics, constructing a gap score function, and taking the edge computing node with the minimum gap score and the maximum resource allocation result as the optimal edge computing node. The dual optimization mechanism enables the system to fully utilize the success experience of historical tasks while ensuring low delay and high bandwidth, more quickly select the most suitable node for resource allocation, and improve the transmission quality and task stability.

Description

Technical Field

[0001] The present invention relates to the technical field of conference systems, and in particular to a paperless conference system based on big data. Background Art

[0002] With the rapid development of information technology and the enhancement of environmental awareness, the traditional paper-based conference model has gradually exposed many shortcomings, such as high cost, low efficiency, and environmental protection. The paperless conference system based on big data technology has emerged. It can not only solve the above problems, but also further improve the organizational efficiency and service quality of the conference through intelligent data processing and analysis. The paperless conference system based on big data comprehensively uses a variety of advanced technologies such as big data, cloud computing, mobile Internet and artificial intelligence, aiming to build an efficient, safe and environmentally friendly modern conference platform. With the continuous advancement and improvement of these technologies, the future paperless conference system will be more intelligent and convenient, providing users with a better conference experience.

[0003] In modern paperless conference systems, the selection of edge computing nodes is crucial to ensure high-quality real-time communication and efficient task execution. In particular, the analysis report generated by historical meetings is one of the key steps to ensure efficient task execution. Specifically, the analysis report is used to transmit to the edge computing node and the task scheduling module. According to the task requirement information in the analysis report, the resource configuration is adjusted to ensure that the task can be executed efficiently and stably, so as to provide optimization suggestions for future meetings. Traditional resource allocation strategies usually give priority to the delay and bandwidth of the node, because these two factors are directly related to the quality of network transmission and user experience. Usually, the delay and bandwidth of the node are given priority. Selecting the node closest to the terminal device can reduce the delay, while selecting the node with larger bandwidth can better meet the needs of real-time data transmission, while ignoring other important network quality indicators. As a result, although the selected node has low delay and large bandwidth, the actual transmission quality is not ideal, which affects the stability of the task. Summary of the invention

[0004] The purpose of the present invention is to provide a paperless conference system based on big data, comprising the following modules:

[0005] Conference task division module: used to obtain conference tasks and divide them into several subtasks; subtasks include real-time tasks and non-real-time tasks;

[0006] Subtask data analysis module: connected with the conference task division module, used to obtain and analyze subtask data, and generate an analysis report containing real-time task demand information, non-real-time task demand information, and participant locations based on the analysis results;

[0007] Subtask scheduling module: connected to the subtask data analysis module, used to obtain all edge computing nodes, filter out the optimal edge computing nodes based on all edge computing nodes, and perform resource transmission for the analysis report;

[0008] The task requirement information in the analysis report is used as a reference requirement; the status information of each edge computing node, including: the network delay, bandwidth, physical distance between the participant's location and the edge computing node, and load of each edge computing node;

[0009] Preset the transmission path between the participant's location and the edge computing node, obtain the packet loss rate, number of hops, and physical distance on the transmission path; obtain the bandwidth of the edge computing node; calculate the connection distance between the participant's location and the edge computing node based on the packet loss rate, number of hops, physical distance, and bandwidth of the edge computing node on the transmission path;

[0010] Obtain key features and non-key features of historical subtasks, divide the features of each subtask into key features and non-key features, calculate key feature similarity based on key features of the current subtask and key features of historical subtasks; calculate non-key feature similarity based on non-key features of the current subtask and non-key features of historical subtasks; calculate subtask similarity based on key feature similarity and non-key feature similarity;

[0011] The resource allocation result of the subtask is calculated based on the connection distance, subtask similarity, and the status information of each edge computing node. The optimal edge computing node is selected from the edge computing nodes based on the resource allocation result of the subtask and the reference requirements. The analysis report of the subtask is assigned to the optimal edge computing node, and a transmission instruction is generated.

[0012] Central Processing Unit: connected to the subtask scheduling module, used to receive and parse the transmission instructions of the subtask scheduling module, and process and summarize the analysis report transmitted by the subtask scheduling module;

[0013] Optimization suggestion module: connected to the central processor, used to respond to the request of the central processor and optimize the meeting content based on the processed and summarized analysis report.

[0014] In another embodiment, the reference requirements include the following status information: network delay, bandwidth, distance requirement, and subtask similarity required by each subtask.

[0015] In another implementation, for real-time tasks:

[0016] The network delay, bandwidth, and connection distance of each edge computing node are obtained, and the real-time task resource allocation results of each edge computing node are calculated.

[0017] In another implementation, for non-real-time tasks:

[0018] Obtain the load, bandwidth, and subtask similarity of each edge computing node, and calculate the non-real-time task resource allocation result of each edge computing node.

[0019] In another embodiment, the real-time task resource allocation result of each edge computing node is calculated by the following formula:

[0020]

[0021] Among them, S i is the real-time task resource allocation result of the i-th edge computing node, L i is the connection distance between the ith edge computing node and the participant, is the network delay of the ith edge computing node, B i is the bandwidth of the i-th edge computing node, α is the adjustment parameter of the bandwidth of the edge computing node, ω1, ω2, ω3 are weight coefficients respectively, and n is the total number of edge computing nodes.

[0022] In another embodiment, the non-real-time task resource allocation result of each edge computing node is calculated by the following formula:

[0023]

[0024] Among them, H i is the non-real-time task resource allocation result of the i-th edge computing node, X is the subtask similarity, Q i is the load of the i-th edge computing node, ω4, ω5, and ω6 are weight coefficients respectively.

[0025] In another embodiment, the connection distance is:

[0026]

[0027] Among them, d i is the physical distance between the ith edge computing node and the participant’s location, h i is the number of hops from the participant’s location to the i-th edge computing node, q i is the packet loss rate of the transmission path between the participant’s location and the i-th edge computing node, β is the adjustment coefficient of the physical distance, ∈ is the minimum value, e is the exponential function, and log is the logarithmic function with base 10.

[0028] In another embodiment, the subtask similarity is:

[0029]

[0030] Among them, x1 is the eigenvalue corresponding to the key feature of the current subtask, y1 is the eigenvalue corresponding to the key feature of the historical subtask, and x j is the feature value corresponding to the jth non-critical feature of the current subtask, y j is the eigenvalue corresponding to the jth non-critical feature of the historical subtask, γ is the adjustment coefficient, δ j is the weight of the jth non-critical feature, and m is the total number of non-critical features in the subtask.

[0031] In another embodiment, selecting the optimal edge computing node includes:

[0032] According to the status information of each edge computing node and the status information of the reference demand, a gap scoring function is constructed to calculate the gap score between each edge computing node and the reference demand, and the edge computing node with the smallest gap score and the largest resource allocation result is selected as the optimal edge computing node;

[0033] The gap score function is:

[0034]

[0035] Among them, D ref (i) Score the gap between the i-th edge computing node and the reference demand, is the kth state information of the ith edge computing node, r m is the status information of the mth reference demand, z is the total number of edge computing node status information, and p is the total number of reference demand status information.

[0036] Technical effects and advantages of the present invention:

[0037] Based on technical defects, this application focuses on the connection distance and subtask similarity to select the optimal edge computing node on the basis of existing technologies. It not only comprehensively evaluates the physical distance between participants and edge computing nodes, number of hops, bandwidth, packet loss rate and delay and other network quality indicators, but also introduces the task similarity calculation resource scheduling results based on historical task data to ensure that the current task can match the most suitable edge computing node used by the historical task. Finally, according to the status information of each edge computing node and the characteristics of the reference requirements, a gap scoring function is constructed, and the edge computing node with the smallest gap score and the largest resource allocation result is used as the optimal edge computing node. This dual optimization mechanism enables the system to fully utilize the successful experience of historical tasks while ensuring low latency and high bandwidth, and select the most suitable node for resource allocation more quickly, thereby improving transmission quality and task stability. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1A schematic diagram of the structure of a paperless conference system based on big data provided in this embodiment. DETAILED DESCRIPTION

[0039] The following will be combined with the drawings in the embodiments of the present invention to clarify the technical solutions in the embodiments of the present invention; it is clearly described in detail. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0040] like Figure 1 As shown, this embodiment discloses a paperless conference system based on big data, including the following modules:

[0041] Conference task division module: used to obtain conference tasks and divide them into several subtasks; subtasks include real-time tasks and non-real-time tasks.

[0042] First, obtain complete meeting task information from the meeting management system or user interface. Meeting task information includes: meeting type (video conference, online collaboration, file sharing, etc.), participant list, meeting schedule, meeting content (presentation, video stream, audio stream, etc.), task priority, special requirements (such as high-definition video, low latency requirements, etc.).

[0043] The system automatically identifies and divides different subtasks according to the nature and needs of the meeting tasks. Specifically, the system divides tasks into real-time tasks and non-real-time tasks according to the real-time requirements of the tasks. Real-time task data includes: video stream, audio stream, real-time collaboration data, real-time control instructions; non-real-time task data includes: file transfer, data processing and analysis, storage and archiving, batch processing tasks.

[0044] Subtask data analysis module: connected to the conference task division module, used to obtain and analyze subtask data, and generate an analysis report containing real-time task requirement information, non-real-time task requirement information, and participant locations based on the analysis results.

[0045] The subtask data analysis module uses existing machine learning or deep learning models to analyze and predict subtask data, predict future real-time task data and non-real-time task data, and generate analysis reports based on task data.

[0046] Subtask scheduling module: connected to the subtask data analysis module, used to obtain all edge computing nodes, filter out the optimal edge computing nodes based on all edge computing nodes, and perform resource transmission for the analysis report.

[0047] The task requirement information in the analysis report is used as the reference requirement; the reference requirement includes the following status information: network delay, bandwidth, distance requirements, and subtask similarity required for each subtask. The status information of each edge computing node includes: network delay, bandwidth, physical distance between the participant's location and the edge computing node, and load of each edge computing node.

[0048] The system uses a routing algorithm to preset the transmission path between the participant's location and the edge computing node based on the geographic location of the participant and the edge computing node. Obtain the packet loss rate, number of hops, and physical distance on the transmission path. The system periodically sends ICMP to each intermediate node on the transmission path and counts the returned response packets. Calculate the packet loss rate by comparing the number of packets sent and received. Obtain the number of hops on the transmission path through the Traceroute tool. Obtain the physical distance on the transmission path through GIS. Obtain the bandwidth of the edge computing node through the edge computing node management platform. Calculate the connection distance between the participant's location and the edge computing node based on the packet loss rate, number of hops, physical distance, and bandwidth of the edge computing node on the transmission path.

[0049]

[0050] Among them, d i is the physical distance between the ith edge computing node and the participant’s location, h i is the number of hops from the participant’s location to the i-th edge computing node, q i is the packet loss rate of the transmission path between the participant's location and the i-th edge computing node, β is the adjustment coefficient of the physical distance, ∈ is the minimum value, ∈=10 -6 , to avoid the denominator being zero, e is the exponential function, and log is the logarithmic function with base 10.

[0051] By calculating the connection distance through the above formula, the network quality between the participant's location and the edge computing node can be comprehensively evaluated, thus providing a scientific basis for task scheduling. This formula can more accurately reflect the actual quality of network transmission by weighting key indicators such as physical distance, number of hops, and packet loss rate, and introducing exponential functions and logarithmic functions for nonlinear adjustment. Specifically, the physical distance and number of hops are attenuated by the exponential function, and the packet loss rate is adjusted by the logarithmic function, ensuring that the connection distance can be significantly increased in the case of poor network quality, thereby avoiding the selection of poor quality transmission paths.

[0052] Obtain key features and non-key features of historical subtasks; divide the features of each subtask into key features and non-key features, and calculate the key feature similarity based on the key features of the current subtask and the key features of historical subtasks; calculate the non-key feature similarity based on the non-key features of the current subtask and the non-key features of historical subtasks; calculate the subtask similarity based on the key feature similarity and the non-key feature similarity.

[0053] The system obtains all completed subtask data from the historical task database. This data includes but is not limited to:

[0054] Subtask type: real-time tasks (such as video streaming, audio streaming, real-time collaboration) or non-real-time tasks (such as file transfer, data processing, storage and archiving);

[0055] Task execution time: the start time and end time of the subtask;

[0056] Task execution result: whether the task was completed successfully, and whether there were any abnormal situations (such as freeze, packet loss, timeout, etc.);

[0057] Network quality indicators: packet loss rate, number of hops, physical distance, bandwidth, etc. on the transmission path;

[0058] Resource usage: edge computing node load, storage space, CPU utilization, etc.

[0059] Participant information: participant's geographic location, device type, network status, etc.

[0060] Key features refer to features that have a direct impact on the results of task execution and are usually closely related to the success of the task. Exemplarily, the key features of real-time tasks are: video resolution, frame rate, encoding format, audio sampling rate, delay requirements, jitter, interaction frequency and response time of real-time collaboration, delay requirements of real-time control instructions, and operation response time. Non-key features of real-time tasks are: device type of participants, operating system version of participants, and network operator of participants. Key features of non-real-time tasks are: file size, transmission priority, complexity of data processing, estimated processing time, storage space requirements, backup strategy, execution time of batch tasks, and task priority. Non-key features of non-real-time tasks are: file creation time, data processing start time, storage file naming rules, and batch task logging.

[0061] Finally, obtain the feature values ​​of the key features and non-key features of the above historical subtasks and the current subtask, and calculate the task similarity:

[0062]

[0063] Among them, x1 is the eigenvalue corresponding to the key feature of the current subtask, y1 is the eigenvalue corresponding to the key feature of the historical subtask, and x j is the feature value corresponding to the jth non-critical feature of the current subtask, y j is the eigenvalue corresponding to the jth non-critical feature of the historical subtask, γ is the adjustment coefficient, δ j is the weight of the jth non-critical feature, and m is the total number of non-critical features in the subtask.

[0064] By statistically analyzing historical task data, the impact of each non-critical feature on the task execution results is analyzed to determine the weight of the non-critical feature. The greater the impact, the greater the weight of the non-critical feature.

[0065] By calculating the task similarity through the above formula, the key feature values ​​and non-key feature values ​​of historical subtasks and current subtasks can be comprehensively considered, so as to more comprehensively evaluate the similarity between the current task and historical tasks. This calculation method can effectively distinguish the criticality and non-criticality of tasks, ensuring that when scheduling tasks and selecting nodes, key features that have a direct impact on the results of task execution are given priority, while taking into account the impact of non-critical features. This provides a scientific basis for selecting nodes, ensuring that tasks can be executed efficiently and stably, especially when network conditions are complex and changeable, and can select the optimal edge computing node to improve the success rate of tasks and user experience.

[0066] The resource allocation results of the subtasks are calculated based on the connection distance, subtask similarity, and the status information of each edge computing node. Specifically:

[0067] For real-time tasks:

[0068] The network delay, bandwidth, and connection distance of each edge computing node are obtained, and the real-time task resource allocation results of each edge computing node are calculated.

[0069]

[0070] Among them, S i is the real-time task resource allocation result of the i-th edge computing node, L i is the connection distance between the ith edge computing node and the participant, is the network delay of the ith edge computing node, B i is the bandwidth of the i-th edge computing node, α is the adjustment parameter of the bandwidth of the edge computing node, ω1, ω2, ω3 are weight coefficients respectively, and n is the total number of edge computing nodes.

[0071] By designing different resource allocation calculation formulas for real-time tasks and non-real-time tasks, the system can more accurately match edge computing nodes with task requirements, thereby significantly improving the efficiency and quality of task execution. For real-time tasks, the system focuses on network latency, bandwidth, and connection distance to ensure low latency and high bandwidth support to ensure the smoothness and stability of real-time data transmission. Specifically, the resource allocation results of real-time tasks are calculated by comprehensively considering the connection distance, network latency, and bandwidth of edge computing nodes. This design ensures that when the system selects edge computing nodes, it gives priority to those nodes that can provide the lowest latency and highest bandwidth, especially when the network conditions are complex and changeable, which can effectively avoid jamming and packet loss and improve user experience.

[0072] For non-real-time tasks:

[0073] Obtain the load, bandwidth, and subtask similarity of each edge computing node, and calculate the non-real-time task resource allocation result of each edge computing node.

[0074]

[0075] Among them, H i is the non-real-time task resource allocation result of the i-th edge computing node, X is the subtask similarity, Q i is the load of the i-th edge computing node, ω4, ω5, and ω6 are weight coefficients respectively.

[0076] For non-real-time tasks, the system pays more attention to load, bandwidth, and subtask similarity to ensure the effective use of resources and the final completion quality of the task. The resource allocation results of non-real-time tasks are calculated by comprehensively considering the subtask similarity, load, and bandwidth of the edge computing nodes. This design enables the system to make full use of idle resources and optimize the execution of non-real-time tasks without affecting real-time tasks. For example, the system can select nodes with low load and sufficient bandwidth to handle file transfer or data processing tasks, ensuring that the tasks can be completed efficiently while avoiding occupying too many network resources.

[0077] In this way, the system can flexibly adjust the resource allocation strategy according to the characteristics of different tasks, ensuring that the low latency and high bandwidth requirements of real-time tasks are met, while maximizing the resource utilization of non-real-time tasks. This not only improves the overall performance of the system, but also enhances the flexibility and adaptability of the system, and can better cope with different types of task requirements. Ultimately, this differentiated design provides a scientific basis for task scheduling, ensuring that each task can be efficiently executed on the most appropriate edge computing node, improving the stability of the conference system and user experience.

[0078] According to the resource allocation results and reference requirements of the subtasks, the optimal edge computing node is selected from the edge computing nodes; specifically, according to the status information of each edge computing node and the status information of the reference requirement, a gap scoring function is constructed to calculate the gap score between each edge computing node and the reference requirement, and the edge computing node with the smallest gap score and the largest resource allocation result is selected as the optimal edge computing node;

[0079] The gap score function is:

[0080]

[0081] Among them, D ref (i) Score the gap between the i-th edge computing node and the reference demand, is the kth state information of the ith edge computing node, r m is the status information of the mth reference demand, z is the total number of edge computing node status information, and p is the total number of reference demand status information.

[0082] By constructing a gap scoring function, the system can quantify the gap between each edge computing node and the reference demand, and screen out the optimal edge computing node in combination with the resource allocation results. Specifically, the gap scoring function calculates the gap between each edge computing node and the ideal state by comparing the state information of the edge computing node (such as network latency, bandwidth, load, etc.) with the state information of the reference demand (such as expected low latency, high bandwidth, low load, etc.). The smaller the gap score, the closer the node is to the reference demand and the better it can meet the execution requirements of the task. At the same time, the system will also consider the resource allocation results and select those nodes with not only the smallest gap score but also the largest resource allocation results as the optimal nodes. This comprehensive evaluation method ensures that when the system selects a node, it not only pays attention to whether the current state of the node meets the task requirements, but also considers the resource utilization efficiency of the node and the success rate of task execution.

[0083] Finally, the analysis report of the subtask is assigned to the optimal edge computing node, and a transmission instruction is generated. The transmission instruction includes: subtask ID, subtask type, task data, optimal edge computing node ID, transmission protocol, and priority.

[0084] Central processing unit: connected to the subtask scheduling module, used to receive and parse the transmission instructions of the subtask scheduling module, and process and summarize the analysis report transmitted by the subtask scheduling module.

[0085] Optimization suggestion module: connected to the central processor, used to respond to the request of the central processor and optimize the meeting content based on the processed and summarized analysis report.

[0086] Before the next meeting, based on the content of the analysis report, timely adjust various data of the current meeting task, such as network settings, hardware equipment and corresponding status information, to achieve meeting optimization. Feedback the optimized meeting status and effect to the administrator or participants to ensure that they can understand the improvement of the meeting in a timely manner.

[0087] These measures not only improve the system's automation level, but also enhance the system's flexibility and adaptability, enabling it to better cope with complex and ever-changing conference requirements. By predicting future task requirements in advance, the system can take preventive measures before problems occur, avoiding resource shortages or performance bottlenecks during task execution, thereby improving user experience and conference quality.

[0088] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and alterations may be made to the embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A paperless conference system based on big data, characterized in that: Includes the following modules: Conference task division module: used to obtain conference tasks and divide them into several subtasks; subtasks include real-time tasks and non-real-time tasks; Subtask data analysis module: connected with the conference task division module, used to obtain and analyze subtask data, and generate an analysis report containing real-time task demand information, non-real-time task demand information, and participant locations based on the analysis results; Subtask scheduling module: connected to the subtask data analysis module, used to obtain all edge computing nodes, filter out the optimal edge computing nodes based on all edge computing nodes, and perform resource transmission for the analysis report; Use the task requirement information in the analysis report as reference requirements; Status information of each edge computing node, including: network latency, bandwidth, physical distance between the participant location and the edge computing node, and load of each edge computing node; Preset the transmission path between the participant's location and the edge computing node, obtain the packet loss rate, number of hops, and physical distance on the transmission path; obtain the bandwidth of the edge computing node; calculate the connection distance between the participant's location and the edge computing node based on the packet loss rate, number of hops, physical distance, and bandwidth of the edge computing node on the transmission path; Obtain key features and non-key features of historical subtasks, divide the features of each subtask into key features and non-key features, calculate key feature similarity based on key features of the current subtask and key features of historical subtasks; calculate non-key feature similarity based on non-key features of the current subtask and non-key features of historical subtasks; calculate subtask similarity based on key feature similarity and non-key feature similarity; The resource allocation result of the subtask is calculated based on the connection distance, subtask similarity, and the status information of each edge computing node. The optimal edge computing node is selected from the edge computing nodes based on the resource allocation result of the subtask and the reference requirements. The analysis report of the subtask is assigned to the optimal edge computing node, and a transmission instruction is generated. Central Processing Unit: connected to the subtask scheduling module, used to receive and parse the transmission instructions of the subtask scheduling module, and process and summarize the analysis report transmitted by the subtask scheduling module; Optimization suggestion module: connected to the central processor, used to respond to the request of the central processor and optimize the meeting content based on the processed and summarized analysis report.

2. A paperless conference system based on big data according to claim 1, characterized in that: The reference requirements include the following status information: network latency, bandwidth, distance requirements, and subtask similarity required for each subtask.

3. A paperless conference system based on big data according to claim 2, characterized in that: For real-time tasks: The network delay, bandwidth, and connection distance of each edge computing node are obtained, and the real-time task resource allocation results of each edge computing node are calculated.

4. A paperless conference system based on big data according to claim 3, characterized in that: For non-real-time tasks: Obtain the load, bandwidth, and subtask similarity of each edge computing node, and calculate the non-real-time task resource allocation result of each edge computing node.

5. A paperless conference system based on big data according to claim 4, characterized in that: The real-time task resource allocation result of each edge computing node is calculated by the following formula: Among them, S i is the real-time task resource allocation result of the i-th edge computing node, L i is the connection distance between the ith edge computing node and the participant, is the network delay of the ith edge computing node, B i is the bandwidth of the i-th edge computing node, α is the adjustment parameter of the bandwidth of the edge computing node, ω1, ω2, ω3 are weight coefficients respectively, and n is the total number of edge computing nodes.

6. A paperless conference system based on big data according to claim 5, characterized in that: The non-real-time task resource allocation result of each edge computing node is calculated by the following formula: Among them, H i is the non-real-time task resource allocation result of the i-th edge computing node, X is the subtask similarity, Q i is the load of the i-th edge computing node, ω4, ω5, and ω6 are weight coefficients respectively.

7. A paperless conference system based on big data according to claim 6, characterized in that: The connection distance is: Among them, d i is the physical distance between the ith edge computing node and the participant’s location, h i is the number of hops from the participant’s location to the i-th edge computing node, q i is the packet loss rate of the transmission path between the participant’s location and the i-th edge computing node, β is the adjustment coefficient of the physical distance, ∈ is the minimum value, e is the exponential function, and log is the logarithmic function with base 10.

8. A paperless conference system based on big data according to claim 7, characterized in that: The subtask similarity is: Among them, x1 is the eigenvalue corresponding to the key feature of the current subtask, y1 is the eigenvalue corresponding to the key feature of the historical subtask, and x j is the feature value corresponding to the jth non-critical feature of the current subtask, y j is the eigenvalue corresponding to the jth non-critical feature of the historical subtask, γ is the adjustment coefficient, δ j is the weight of the jth non-critical feature, and m is the total number of non-critical features in the subtask.

9. A paperless conference system based on big data according to claim 8, characterized in that: Screening the best edge computing nodes, including: According to the status information of each edge computing node and the status information of the reference demand, a gap scoring function is constructed to calculate the gap score between each edge computing node and the reference demand, and the edge computing node with the smallest gap score and the largest resource allocation result is selected as the optimal edge computing node; The gap score function is: Among them, D ref (i) Score the gap between the i-th edge computing node and the reference demand, is the kth state information of the ith edge computing node, r m is the status information of the mth reference demand, z is the total number of edge computing node status information, and p is the total number of reference demand status information.

Images