Enterprise series intelligent office collaboration data processing system based on edge computing

Abstract

The invention relates to the technical field of enterprise office collaboration, and discloses an enterprise series intelligent office collaboration data processing system based on edge computing. The system comprises a collaborative data acquisition module, a collaborative feature analysis module, an edge task scheduling module and a dynamic storage optimization module. The collaborative data acquisition module is deployed at a local edge node of an enterprise, captures a collaborative operation behavior flow of the multi-source office terminal in real time, and extracts space-time distribution features to generate an original collaborative feature vector; the collaborative feature analysis module analyzes the coupling strength of related data through a multi-dimensional correlation analysis engine, quantifies the frequency fluctuation amplitude of a communication session, and fuses to generate a collaborative efficiency quantized value; the edge task scheduling module dynamically divides task load levels based on the value, and matches execution priorities to generate a distributed calculation task allocation strategy; and the dynamic storage optimization module responds to the strategy, monitors the throughput delay of the storage nodes, reconstructs a high-frequency access data storage index, and generates a cross-node storage topology configuration.

Description

Technical Field

[0001] This invention relates to the field of enterprise office collaboration technology, specifically to an enterprise-level intelligent office collaboration data processing system based on edge computing. Background Technology

[0002] As enterprises deepen their digital transformation, the amount of collaborative data generated by multiple terminals in office scenarios is exploding, covering various operations such as document editing, instant messaging, and task scheduling. Traditional centralized data processing architectures are increasingly revealing their limitations when dealing with such scenarios. Centralized architectures rely on cloud servers for unified data collection and analysis, resulting in excessively long data transmission paths, susceptibility to network bandwidth fluctuations, and difficulty in meeting the low-latency requirements of real-time collaboration. For example, in cross-departmental document collaboration, the real-time synchronization of editing progress depends on cloud forwarding, often leading to delayed operation feedback and impacting team collaboration efficiency. Meanwhile, centralized architectures concentrate all computing tasks in the cloud. When the amount of collaborative data generated by enterprise edge nodes surges, cloud computing resources can easily become a bottleneck, leading to increased task response latency. In addition, under centralized storage mode, frequently accessed collaborative data needs to be repeatedly retrieved from the cloud, which not only exacerbates network transmission pressure but also makes it impossible to dynamically adjust storage strategies according to data access frequency due to the fixed storage location, resulting in inefficient use of storage resources. Existing collaborative data processing solutions lack sufficient analysis of the correlations between multi-source behavioral data, often processing data from document editing, communication sessions, and task scheduling in a fragmented manner, making it difficult to capture the inherent coupling relationships between different operational behaviors. For example, changes in task scheduling instructions may directly affect the frequency and content of document editing, but existing systems lack the ability to quantitatively analyze such correlations, resulting in a disconnect between generated collaborative strategies and actual office scenario requirements. Furthermore, the computing and storage resources of edge nodes are not fully utilized; in most cases, they are only used as data acquisition terminals and do not participate in local data processing, leading to resource idleness and waste. Summary of the Invention

[0003] The purpose of this invention is to provide an enterprise-level intelligent office collaborative data processing system based on edge computing to solve the problems mentioned in the background art.

[0004] To achieve the above objectives, the present invention provides an enterprise-level intelligent office collaborative data processing system based on edge computing, the system comprising: The collaborative data acquisition module is deployed on the enterprise's local edge node to capture collaborative operation behavior streams generated by multiple source office terminals in real time. The collaborative operation behavior streams include document editing trajectories, communication session frequencies, and task scheduling instruction sequences. The module extracts the spatiotemporal distribution features of the operation behaviors to generate original collaborative feature vectors. The collaborative feature analysis module receives the original collaborative feature vector, analyzes the coupling strength between the document editing trajectory and the task scheduling instruction sequence through a multi-dimensional correlation analysis engine, quantifies the fluctuation range of the communication session frequency within a preset time window, and integrates the coupling strength and fluctuation range to generate a quantified value of collaborative effectiveness. The edge task scheduling module dynamically divides the computing task load level based on the quantified value of the collaborative efficiency, matches the task execution priority according to the real-time resource utilization rate of the enterprise edge nodes, and generates a distributed computing task allocation strategy. The dynamic storage optimization module responds to the distributed computing task allocation strategy, monitors the data throughput latency of edge storage nodes, reconstructs the storage location index of frequently accessed data, and generates a data storage topology configuration across edge nodes.

[0005] Preferably, the system further includes: The collaborative behavior monitoring module, based on the data storage topology configuration, tracks the trend of collaborative operation behavior flow changes of multi-source office terminals, identifies abnormal behavior characteristics that deviate from the preset collaborative mode, and generates a collaborative abnormal feature identifier set. The security response control module calls the collaborative abnormal feature identifier set, blocks the communication session channel associated with abnormal behavior features, resets the execution permission threshold of the task scheduling instruction sequence, and outputs the edge computing resource isolation strategy. The strategy iteration update module integrates the edge computing resource isolation strategy and the collaborative effectiveness quantification value, corrects the coupling strength calculation parameters of the multidimensional correlation analysis engine, and updates the spatiotemporal distribution feature extraction rules of the collaborative data acquisition module.

[0006] Preferably, the original collaborative feature vector includes the version iteration density of the document editing trajectory, the temporal distribution matrix of the communication session frequency, and the concurrent execution volume of the task scheduling instruction sequence; The quantifiable value of the synergistic effect includes the task coupling strength coefficient, communication fluctuation index, and resource matching score. The distributed computing task allocation strategy includes load balancing weights, task priority queues, and edge node resource allocation ratios. The data storage topology configuration defines the storage node location mapping table, data access path rules, and cross-node data synchronization frequency.

[0007] Preferably, the collaborative feature analysis module includes: The behavior coupling parsing submodule parses the dependency chain between the document editing trajectory and the task scheduling instruction sequence, calculates the overlapping execution frequency of key operation nodes in the dependency chain, and generates a task collaboration tightness index. The communication fluctuation monitoring submodule calculates the rate of change of the standard deviation of the communication session frequency within a continuous time unit, detects abnormal communication periods that exceed the preset fluctuation threshold, and generates a communication stability assessment matrix. The efficiency fusion calculation submodule aggregates the task collaboration tightness index and the communication stability evaluation matrix, generates a collaborative efficiency quantification value through weighted normalization processing, and outputs it to the edge task scheduling module.

[0008] Preferably, the edge task scheduling module includes: The load grading engine divides tasks into three levels—lightweight, medium-load, and high-load—based on the numerical range of collaborative performance quantification values, and correlates them with the real-time memory usage and CPU utilization of enterprise edge nodes. The priority matching unit calculates the preemption weight of different task levels on the enterprise edge node computing resources and generates a list of task execution priorities sorted by preemption weight; The allocation strategy generator dynamically adjusts the upper limit of task concurrency processing of enterprise edge nodes based on the task execution priority list, and generates a distributed computing task allocation strategy that includes node load allocation ratio and task queue scheduling sequence.

[0009] Preferably, the dynamic storage optimization module includes: The throughput monitoring unit continuously collects the data read / write response time and network transmission latency of edge storage nodes, and marks inefficient storage nodes whose response time exceeds a preset threshold. The index reconstruction engine analyzes the distribution density of frequently accessed data on inefficient storage nodes, migrates high-density data to nearby enterprise edge nodes, and rebuilds location index labels. The topology configuration generator generates a data storage topology configuration that includes cross-node data routing rules and a list of storage node health statuses, based on the reconstructed location index labels.

[0010] Preferably, the collaborative behavior monitoring module performs the following: Receive the list of storage node health statuses in the data storage topology configuration and monitor the collaborative operation flow of multi-source office terminals on healthy storage nodes; Compare the differences between the collaborative operation behavior flow and the preset collaborative mode in three dimensions: document editing trajectory deviation, communication session frequency deviation, and task scheduling sequence abnormal interruption rate. When the difference in any dimension exceeds the safety threshold, the abnormal behavior features are marked and a collaborative abnormal feature identifier set containing the abnormal type code and time stamp is generated.

[0011] Preferably, the security response control module includes: The session blocking unit parses the anomaly type code in the collaborative anomaly feature identifier set, freezes the transmission bandwidth of the corresponding communication session channel, and generates a channel isolation log. The permission reset engine reduces the maximum concurrent execution of associated task scheduling instruction sequences and sets new permission verification trigger conditions based on the severity level of the channel isolation logs. The resource isolation policy generator integrates channel isolation logs and permission verification trigger conditions to generate edge computing resource isolation policies that include a list of restricted nodes and a set of resource access rules.

[0012] Preferably, the strategy iteration update module executes: Extract the set of resource access rules from the edge computing resource isolation strategy and adjust the weight of the resource matching score calculation for the quantified value of collaborative effectiveness. Based on the adjusted resource matching score, the algorithm parameters of the task collaboration tightness index of the behavior coupling parsing submodule are corrected. The spatiotemporal distribution feature extraction rules of the collaborative data acquisition module are updated synchronously to increase the sampling density during periods of abnormal communication session frequency.

[0013] Preferably, the system further includes: The cross-domain audit tracing module collects the operation records of the collaborative anomaly feature identifier set and the edge computing resource isolation strategy, and generates a cross-domain operation tracing report by associating the audit logs of multiple enterprise edge nodes; The cross-domain operation tracing report input strategy iterative update module triggers the multi-dimensional correlation analysis engine parameter update of the collaborative feature analysis module.

[0014] Preferably, the collaborative data acquisition module captures the collaborative operation behavior flow through lightweight probes deployed at edge nodes; The data storage topology configuration of the dynamic storage optimization module is fed back to the edge task scheduling module, triggering the reconstruction of the distributed computing task allocation strategy; The edge computing resource isolation strategy of the security response control module is synchronized to the collaborative behavior monitoring module, and the security threshold judgment benchmark of the preset collaborative mode is updated.

[0015] Compared with the prior art, the beneficial effects of the present invention are: By deploying collaborative data acquisition modules on enterprise local edge nodes, collaborative operation behavior flows from multiple source office terminals can be captured in real time, and the spatiotemporal distribution characteristics of the operation behavior can be extracted to generate original collaborative feature vectors. This localized data acquisition method reduces the process of transmitting data to the cloud, reduces the latency impact of network transmission, and enables collaborative data to be quickly captured and processed at the edge, which is more in line with the needs of real-time office collaboration scenarios. After receiving the original collaborative feature vector, the collaborative feature analysis module uses a multi-dimensional correlation analysis engine to analyze the coupling strength between the document editing trajectory and the task scheduling instruction sequence. Simultaneously, it quantifies the fluctuation range of communication session frequency within a preset time window and merges the two to generate a quantifiable value of collaborative effectiveness. This process achieves deep correlation analysis of multi-source collaborative data, breaking the limitations of traditional solutions that process data from different operational behaviors in a fragmented manner. It can more comprehensively reflect the actual state of the office collaboration process, providing a more realistic basis for subsequent task scheduling. The edge task scheduling module dynamically classifies computing task load levels based on collaborative efficiency metrics and matches task execution priorities according to the real-time resource utilization of enterprise edge nodes, generating a distributed computing task allocation strategy. This approach fully utilizes the computing resources of the enterprise's local edge nodes, avoiding the load pressure caused by excessive concentration of computing tasks in the cloud, and enabling computing tasks to be rationally allocated according to the resource status of edge nodes, thus improving the flexibility and adaptability of task processing. The dynamic storage optimization module responds to distributed computing task allocation strategies, monitors the data throughput latency of edge storage nodes, reconstructs the storage location index of frequently accessed data, and generates a data storage topology configuration across edge nodes. By dynamically adjusting the data storage locations of frequently accessed data, it can reduce the path length during data access, lower data throughput latency, improve the utilization efficiency of storage resources, and enable the storage architecture to better adapt to the dynamic changes in data access during collaborative office work. Attached Figure Description

[0016] Figure 1 This is a schematic diagram illustrating the working principle of the enterprise-level intelligent office collaborative data processing system based on edge computing as described in this invention. Figure 2 A flowchart for collaborative behavior monitoring and security response; Figure 3 A flowchart for the collaborative feature analysis module; Figure 4 A flowchart for the edge task scheduling module; Figure 5 This is a flowchart of the safety response control module. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] Please see Figure 1This invention provides an enterprise-level intelligent office collaborative data processing system based on edge computing. The system comprises four main modules that work collaboratively in its core architecture. Specific implementation details are as follows: The collaborative data acquisition module is deployed on the enterprise's local edge nodes, capturing collaborative operation behavior streams from multiple office terminals (such as personal computers, mobile devices, and smart conferencing terminals) in real time through built-in lightweight probes. This behavior stream includes document editing trajectories (recording events such as cursor movement, content addition / deletion, and version saving), communication session frequency (timestamps of instant messages and video conference initiation / end, and participant information), and task scheduling instruction sequences (operation instructions such as task creation, assignment, and completion status changes). The module employs a spatiotemporal window analysis algorithm to process the captured operation behavior streams in real time, extracting their temporal and spatial distribution patterns, such as the concentration of document editing trajectories within specific working periods and the distribution characteristics of communication sessions across office nodes in different geographical regions. Ultimately, it generates a raw collaborative feature vector comprising dimensions such as version iteration density, temporal distribution matrix, and concurrent execution volume.

[0019] The collaborative feature analysis module receives raw collaborative feature vectors from the acquisition module. This module incorporates a multi-dimensional correlation analysis engine, first analyzing the dependency relationship between the document editing trajectory and the task scheduling instruction sequence. By calculating the overlap frequency and causal correlation strength of the event sequences on the time axis, it derives the task coupling strength coefficient. Simultaneously, for communication session frequency data, a sliding time window statistical method is used to calculate the standard deviation and rate of change of session frequency within a preset time period, generating a communication fluctuation index. These coefficients and indices are then weighted and normalized by the performance fusion calculation submodule, and combined with real-time resource occupancy status, a resource matching score is calculated, ultimately outputting a quantitative value of collaborative performance.

[0020] The edge task scheduling module uses a dynamic hierarchical strategy to classify computational task load levels based on the received collaborative efficiency quantification value: tasks with low quantification values ​​are classified as lightweight tasks, those with medium quantification values ​​as medium-load tasks, and those with high quantification values ​​as high-load tasks. The load grading engine combines the real-time CPU utilization and memory usage of the enterprise's edge nodes to calculate the resource preemption weight for different task levels; the priority matching unit generates a sorted list of task execution priorities based on the preemption weights; the allocation strategy generator dynamically adjusts the task concurrency limit of each edge node based on this list, ultimately outputting a distributed computing task allocation strategy that includes load balancing weights, task priority queues, and node resource allocation ratios.

[0021] In response to the aforementioned allocation strategy, the dynamic storage optimization module continuously collects data read / write response times and network transmission latency of edge storage nodes through the throughput monitoring unit, identifying inefficient nodes with response times exceeding thresholds. The index reconstruction engine analyzes the heatmap of high-frequency access data distribution on inefficient nodes, triggering a data migration operation: migrating high-density access data to nearby edge nodes and rebuilding location index labels. The topology configuration generator generates a data storage topology configuration based on the new index labels. This configuration includes a storage node location mapping table, cross-node data routing rules, data synchronization frequency, and a list of node health statuses. This configuration is fed back to the edge task scheduling module, driving the reconstruction of its task allocation strategy.

[0022] Example 1: Reading Figure 2 The collaborative behavior monitoring module operates based on the data storage topology configuration output by the dynamic storage optimization module. This configuration includes a list of storage node health statuses, clearly identifying the set of currently available and performance-compliant storage nodes. This module monitors healthy storage nodes, collecting real-time collaborative operation behavior flow data generated by multi-source office terminals deployed on them. The collected behavior flows cover three main categories: document editing trajectory, communication session frequency, and task scheduling instruction sequence. The module has a built-in preset collaborative mode, established based on historical normal collaborative behavior, defining the baseline parameter range for each behavior flow. The monitoring process continuously performs a three-dimensional comparative analysis: the first dimension calculates the trajectory deviation between the real-time document editing trajectory and the preset mode, quantifying the deviation by analyzing the dispersion of editing operations over time and the anomalies in the distribution of operation types; the second dimension monitors the communication session frequency, comparing the difference between the actual number of sessions initiated per unit time and the preset frequency baseline value to generate a frequency deviation value; the third dimension tracks the task scheduling instruction sequence, calculating the abnormal interruption rate by recording the number of unexpected interruptions and violations of the preset execution sequence chain. Each of the above three dimensions has an independent security threshold, which is flexibly adjusted according to the enterprise's collaborative strategy. When behavioral data flows through the monitoring module, if the document editing trajectory deviation exceeds the trajectory safety threshold, the communication session frequency deviation exceeds the frequency safety threshold, or the task scheduling sequence abnormal interruption rate exceeds the interruption safety threshold, the monitoring module immediately marks the behavioral characteristics of that period as abnormal. The marking process associates precise timestamp information and abnormal dimension identifiers. Simultaneously, the module integrates the abnormal feature type and trigger time point to generate a structured collaborative abnormal feature identifier set. This identifier set includes abnormal type encoding, severity level marking, and time metadata such as the first occurrence and duration.

[0023] Upon receiving the coordinated anomaly feature identifier set, the security response control module initiates a response process. This module includes a session blocking unit, which first parses the anomaly type code in the identifier set. If the code points to a communication session anomaly, the unit locates the source terminal, target terminal, and communication channel identifier of the abnormal communication session. Subsequently, the blocking unit performs a channel freeze operation: limiting the maximum available transmission bandwidth of the channel to a preset minimum guaranteed value and completely blocking new session connection requests. During this operation, a channel isolation log is generated synchronously, recording in detail the unique identifier of the blocked channel, the timestamp of the blocking operation, the applied bandwidth limit value, and the anomaly type association code. Simultaneously, the security response control module activates the permission reset engine, which implements gradient control measures based on the anomaly severity level recorded in the channel isolation log. The permission reset engine first parses the severity level identifier; if the level reaches a preset high-risk standard, the engine reduces the maximum concurrent execution thread count of the associated task scheduling instruction sequence by a factor calculated based on a severity gradient model. The permission reset engine further sets new permission verification trigger conditions, stipulating that when an associated task scheduling instruction meets specific risk characteristics, additional identity authentication verification or approval process chains must be executed before operation can continue. The resource isolation policy generator ultimately integrates key parameters from channel isolation logs with detailed rules for permission verification triggering conditions. The generator establishes a list of restricted nodes, explicitly including terminal devices and edge node identifiers exhibiting abnormal behavior; simultaneously, it generates a set of resource access rules, clearly defining resource access constraints for restricted nodes under specific time periods or operation types. The final output is an edge computing resource isolation policy that integrates the above elements, explicitly specifying the restricted objects, restricted operation types, restricted execution time domains, and verification mechanisms.

[0024] The strategy iteration update module receives the edge computing resource isolation strategy output by the security response control module and the continuously generated collaborative effectiveness quantification dataset within the system. This module performs dynamic parameter and rule iteration operations. First, the module parses the resource access rule set in the edge computing resource isolation strategy, extracting the defined resource usage constraints and risk node resource access characteristics. Based on this information, the strategy iteration update module adjusts the calculation weight factor of the resource matching score sub-item in the collaborative effectiveness quantification calculation model, increasing the influence of the resource usage status of nodes or regions constrained by the resource isolation strategy on the overall score. The adjusted resource matching score further drives the module to execute parameter corrections in the behavior coupling analysis sub-module. The strategy iteration update module analyzes historical collaborative effectiveness quantification data and the new score weights to calculate the optimization adjustment direction of key parameters in the original task collaboration tightness index algorithm in the behavior coupling analysis sub-module. For example, updating the time window analysis parameters to cover a longer behavior association period, or adjusting the event time tolerance threshold parameter in the dependency identification algorithm. These algorithm parameter corrections are dynamically injected into the behavior coupling analysis sub-module of the collaborative feature analysis module. Simultaneously, the strategy iteration update module triggers rule updates in the collaborative data acquisition module based on communication session frequency monitoring results. When the communication fluctuation monitoring submodule continuously reports high-frequency abnormal communication behavior characteristics during a specific period, the update module generates an instruction requiring the collaborative data acquisition module to increase the data sampling frequency and sampling depth within a specified time period. This instruction is formatted as a new spatiotemporal distribution feature extraction rule and injected into the acquisition module, enabling the acquisition module to capture more granular communication session behavior feature sequences during periods of high anomaly incidence. The entire strategy iteration process forms a closed loop: the system dynamically optimizes feature extraction, analysis algorithms, and resource control strength based on security event feedback, continuously adapting to the dynamic changes in the enterprise collaborative environment and the evolution of risk patterns. After receiving the updated edge computing resource isolation policy, the collaborative behavior monitoring module synchronously retrieves relevant rules from its built-in policy library and dynamically updates the judgment benchmark logic of the three-dimensional security threshold in its preset collaborative mode, achieving consistent adjustment of anomaly identification sensitivity and risk response strength.

[0025] Example 2: Reading Figure 3The process of generating the original collaborative feature vector by the collaborative data acquisition module includes three dimensions of structured data encapsulation. For captured document editing trajectory data, the module calculates the product of the number of document state changes and the magnitude of content modifications per unit time, encapsulating this into a version iteration density parameter. This parameter quantifies the dynamic intensity during the document collaborative iteration process. For communication session frequency data, the module divides communication events according to predetermined time units, classifies communication object identifiers, single session duration, and time period type, forming a multi-dimensional time-series distribution matrix. This matrix reflects the patterned spatiotemporal distribution characteristics of communication behavior. For task scheduling instruction sequence data, the module aggregates instruction events generated by different terminals within the same time period, calculates the scale of tasks that can be processed in parallel, and generates a concurrent execution quantity value, which reflects the parallel processing capability of task scheduling.

[0026] The collaborative feature analysis module comprises three interconnected sub-modules that process the aforementioned feature vectors. The behavior coupling analysis sub-module performs correlation analysis on the document editing trajectory and task scheduling instruction sequence. This sub-module establishes an event sequence timeline and identifies the dependency chain structure between trajectory events and instruction events. Dependencies include explicit and implicit dependencies: explicit dependencies are manifested as a sequence of editing operations on associated documents directly triggered after a task state change; implicit dependencies are manifested as dense groups of document saving behaviors appearing before the task deadline. The analysis sub-module calculates the key operation nodes in the dependency chain and counts the frequency of overlapping execution of nodes on the timeline. This frequency value characterizes the collaboration tightness between the task and the document, forming a task collaboration tightness index output.

[0027] The communication fluctuation monitoring submodule processes time-series distribution matrix data. This submodule defines continuous time units as the basic analysis window, and calculates the standard deviation of communication session frequencies within each window. The module continuously calculates the rate of change of the standard deviation between adjacent time windows. The system presets a communication fluctuation threshold parameter; when the rate of change within a continuous time window is consistently higher than this threshold, the corresponding time period is marked as an abnormal communication period. The monitoring submodule integrates the fluctuation score status and abnormal marker distribution of each time period to generate a communication stability assessment matrix reflecting the overall communication stability level.

[0028] The performance fusion calculation submodule receives the task collaboration tightness index from the behavior coupling analysis submodule and the communication stability assessment matrix from the communication fluctuation monitoring submodule. The fusion process employs a weighted normalization algorithm for data aggregation. The algorithm sets a weight coefficient α for the task collaboration tightness index and a weight coefficient β for the communication stability assessment matrix, which are then applied to the original index data for weighted calculation. During the calculation, the real-time resource occupancy status parameters of the enterprise edge nodes are simultaneously referenced, and the resource alignment of the output results is adjusted through a resource matching factor. The final generated collaborative performance quantification value contains three structured numerical components: the task coupling strength coefficient, directly derived from the normalized transformation result of the task collaboration tightness index; the communication fluctuation index, reflecting the integrated value of the distribution density and fluctuation amplitude during abnormal periods in the communication stability assessment matrix; and the resource matching score, reflecting the degree of adaptation between the quantification value and the current node's resource status.

[0029] The load grading engine in the edge task scheduling module classifies tasks based on a combination of the task coupling strength coefficient and communication fluctuation index in the quantified values ​​of collaborative effectiveness. The engine pre-defines a mapping rule for quantified value ranges: tasks with a task coupling strength coefficient above a set upper limit and a communication fluctuation index below a set lower limit are classified as lightweight tasks; both the task coupling strength coefficient and communication fluctuation index are within the median range; and tasks with a task coupling strength coefficient below a set lower limit and a communication fluctuation index above a set upper limit are classified as high-load tasks. This classification is also based on real-time operating parameters of the enterprise's edge nodes, including continuously monitored CPU utilization percentage and memory resource usage. The load grading engine provides a classification basis for subsequent scheduling processes through task level labels.

[0030] The priority matching unit calculates the resource preemption weight model based on the task level labels output by the load grading engine. The unit pre-sets weight gradient allocation rules: lightweight tasks are assigned the lowest preemption weight value, high-load tasks are assigned the highest preemption weight value, and medium-load tasks are assigned a medium weight value between the two. This unit sorts the tasks in descending order of preemption weight value, forming a structured task execution priority list. The allocation strategy generator receives this priority list and dynamically adjusts the task scheduling strategy based on the real-time resource availability status of each edge node. The generator sets a dynamic task concurrency limit for each edge node and generates a distributed computing task allocation strategy based on the task order in the list and the node load status. This strategy specifically includes the proportional relationship of task load distribution between different nodes and the scheduling sequence rules of the execution task queues within each node, forming a complete task scheduling and execution framework.

[0031] Example 3: Reading Figure 4The edge task scheduling module receives the quantified value of collaborative effectiveness as input. The task coupling strength coefficient and communication fluctuation index in the quantified value are parsed and processed by the load grading engine. The load grading engine defines the task level determination rules: when the task coupling strength coefficient is within a set high range and the communication fluctuation index is below a set low range threshold, it is assigned a lightweight task level label; when both the task coupling strength coefficient and the communication fluctuation index are within a set medium range threshold, it is assigned a medium load task level label; when the task coupling strength coefficient is below a set low range threshold and the communication fluctuation index exceeds a set high range threshold, it is assigned a high load task level label. This level classification process is linked in real time to the percentage values ​​of CPU utilization and memory resource occupancy reported by the enterprise edge node resource status monitoring system. The engine dynamically establishes a mapping table between task levels and node resource status, which records the expected load parameters of different task levels under different resource states.

[0032] The priority matching unit calculates the preemption intensity of computing resources by tasks based on the task level identifier output by the load grading engine. This unit configures a gradient preemption weight model based on task level: a lightweight task level allocation of basic preemption weight values. ;Assign weight values ​​to medium-load task levels High-load task level allocation weight value . formula Used to dynamically adjust standard weight values ​​to reflect real-time resource pressure. In the formula This represents the dynamic weighting factor of the current application; This represents the percentage of CPU utilization. This represents the percentage of memory resource utilization. and This is a preset adjustment coefficient used to adjust the relative contributions of processor and memory factors in weight correction. The priority matching unit applies this dynamic weight factor to adjust the standard weight values ​​to generate the final preemption weight, using the following formula: ,in represent , or One of the tasks is to sort them from highest to lowest according to their final preemption weight values ​​and generate a task execution priority list. The list structure includes the task identifier, the final preemption weight value, and a description of the expected resource requirements.

[0033] The allocation strategy generator processes the task execution priority list and real-time snapshots of the operational status of each edge node in the enterprise. The generator calculates the maximum concurrent processing capacity that each edge node can handle, dynamically determined based on the percentage of remaining available computing power of the node's CPU and the total amount of free memory resources. Based on the priority list order and the node's concurrent capacity, the generator constructs a task scheduling sequence using a sequential allocation mechanism. The allocation process follows this logic: tasks with high preemption weights in the task execution priority list are preferentially allocated to the edge nodes with the best resource availability; tasks with the same weight in the task execution priority list are discretized based on the goal of load balancing between nodes. The final output distributed computing task allocation strategy includes two key components: a node load allocation ratio table that clearly specifies the proportional relationship between the different task levels that each edge node should handle; and a task queue scheduling sequence that details the execution order chain and time window planning for specific tasks on each edge node.

[0034] The throughput monitoring unit of the dynamic storage optimization module continuously polls all edge storage nodes registered in the system. The unit collects the data read / write request response time (in milliseconds) and cross-node network transmission round-trip latency (in milliseconds) for each storage node. The system sets read / write response threshold parameters and network latency threshold parameters. The monitoring unit compares the collected data with the threshold parameters: storage nodes whose response time exceeds the read / write response threshold or whose network latency exceeds the network latency threshold are marked as inefficient storage nodes. The monitoring unit generates a performance anomaly report containing a list of inefficient nodes and a corresponding performance status description.

[0035] The index reconstruction engine receives performance anomaly reports and high-frequency access data location records. The engine calculates the total access frequency of high-frequency data on each inefficient storage node within a unit of time and constructs a heatmap based on the access frequency distribution. The engine introduces the concept of a heat index: defining the heat index of a node as... ,in This indicates the number of frequently accessed data objects carried by this node. This indicates the access frequency of data object d at this node, and T represents the length of the statistical time period. The system sets a heat index migration threshold parameter. The engine compares the heat index of each inefficient storage node: nodes with heat indices exceeding the migration threshold trigger data migration operations. The engine selects the migration target node based on its network topology proximity to the source node and the current load status of the target node. After the migration operation is executed, the location index label of the migrated data object is reconstructed. The new label contains the physical address identifier of the target node, access port information, and protocol support types.

[0036] The topology configuration generator integrates the reconstructed set of location index tags and real-time operational monitoring data of edge storage nodes. The generator creates a storage node location mapping table, which records unique identifiers for data objects, current storage node address indexes, and index status flags. The generator defines a set of cross-node data routing rules, detailing the preferred order of data transmission paths between different nodes, backup path triggering conditions, and data transmission protocol type limitations. The generator configures a set of data synchronization frequency parameters, setting different synchronization cycle lengths based on data access frequency levels. The generator maintains a dynamic list of storage node health statuses, integrating storage node availability status, current performance scores, and fault flag information. These elements are integrated to form the final data storage topology configuration. This configuration is transmitted to the edge task scheduling module and the collaborative behavior monitoring module. The edge task scheduling module uses the configured storage node health status list to update the node availability database, dynamically adjusting subsequent distributed computing task allocation strategies; the collaborative behavior monitoring module determines the set of healthy data storage locations based on the storage node location mapping table, thereby focusing on the target deployment locations of monitored resources.

[0037] Example 4: The collaborative behavior monitoring module detected anomalies in the collaborative operation flow of a group of office terminals during periodic scanning. The table below shows a fragment of the generated collaborative anomaly feature set:

[0038] The security response control module parses the identifier set. The session blocking unit identifies high-frequency abnormal communication characteristics corresponding to the C01 type code. This unit locates the communication channel ID used by user A as COM_ChA17. Channel freezing is executed: the maximum bandwidth of this channel is limited to 50Kbps, new session establishment requests are blocked, and a channel isolation log is generated to record the blocked channel parameters and operation time.

[0039] The permission reset engine takes action based on the severity level field in the isolation log. For S3-level anomalies, the engine reduces the maximum concurrency of the associated task sequence: TaskGroup_A, which originally allowed 100 concurrent threads, is limited to a maximum of 20 threads. Simultaneously, a permission verification trigger condition is set: when user A attempts to start a task with a priority > 7, the biometric authentication process is forcibly enabled.

[0040] The resource isolation policy generator integrates the processing results. Based on the spatial distribution of anomaly features, the generator includes the terminal Device_A17 and the edge node Edge_Node05 in the restricted node list. A set of resource access rules is defined: Rule 1: Access to the core storage area by restricted nodes requires approval. Rule 2: High-frequency communication sessions require a 5-second routing delay. Rule 3: Editing of sensitive documents triggers real-time watermarking The final generated edge computing resource isolation policy clearly identifies the restricted objects, the types of restricted operations, and the effective time period. This policy is pushed to the relevant functional modules for execution in real time.

[0041] The strategy iteration and update module receives the resource isolation policy. The module extracts the resource access rule set and identifies rule 3, which involves specific constraints related to document operations. Based on these constraints, the module updates the calculation model for collaborative effectiveness metrics: increasing the weight of the document security operation factor in the resource matching score from the original baseline of 0.15 to 0.28. This weight adjustment affects subsequent effectiveness evaluation results.

[0042] The module synchronously analyzes the historical collaborative effectiveness quantification dataset. It was found that the adjusted resource matching score exhibits a significant changing trend in document-intensive tasks. Based on this, parameter adjustments were triggered for the behavior coupling parsing submodule: the time window analysis parameter of this submodule was extended from the default 300 seconds to 420 seconds, increasing the observation period for the correlation between task sequences and document operations. The event time tolerance threshold in the algorithm kernel was adjusted from ±15 seconds to ±8 seconds, improving the accuracy of operation correlation detection.

[0043] The collaborative data acquisition module receives policy iteration update instructions. When communication fluctuation monitoring records show a continuous anomaly during a specific period (such as three consecutive C01 anomalies), the acquisition module automatically activates an enhanced sampling mode: compressing the communication session frequency sampling interval from 60 seconds to 20 seconds within the marked period; upgrading the document editing trajectory capture depth from recording operation type to recording operation content hash value. New spatiotemporal distribution feature extraction rules are dynamically configured and injected into the acquisition process.

[0044] The system establishes a closed-loop update mechanism. Upon receiving the updated resource isolation policy, the collaborative behavior monitoring module immediately adjusts its anomaly detection criteria: for terminal devices in the restricted node list, the tolerance threshold for document editing trajectory deviation is reduced from 15% to 8%; the benchmark for high-frequency communication sessions is adjusted from 50 times per minute to 30 times per minute. The monitoring module synchronously receives notifications of changes to behavior coupling parameters pushed by the policy iteration update module, dynamically optimizing the expected execution path model of the task scheduling sequence in the preset collaborative mode.

[0045] During implementation, when Device_A17 triggers document editing again outside of working hours, the collaborative behavior monitoring module immediately identifies the anomaly based on the updated judgment criteria. The system automatically embeds a digital watermark according to resource isolation policy rule 3, while the security response control module records the characteristics of this violation. This characteristic data is input into the policy iteration update module, triggering the next round of parameter adjustment: the document security operation factor weight is further increased to 0.35, and the time window parameter is expanded to 480 seconds. The entire process forms a continuously optimized adaptive protection mechanism, with each module achieving policy linkage and parameter co-evolution through structured data transmission.

[0046] Example 5: The cross-domain audit tracing module continuously collects collaborative anomaly feature identifier sets, edge computing resource isolation policy execution records, and raw audit log streams distributed across multiple enterprise edge nodes. This module establishes a unified timeline alignment mechanism, calibrating the timestamp precision of all input data to the millisecond level. The audit log streams contain heterogeneous data such as end-user operation events, system-triggered task scheduling instruction change records, and storage node access trajectories. The module's core operation link tracing function establishes cross-node associations based on two key identifiers: first, the transaction identifier of the anomaly event, which is embedded when the collaborative anomaly feature identifier set is generated; and second, the execution batch code of the resource isolation policy, which runs through the entire lifecycle of policy implementation.

[0047] When new data fragments are acquired, the module performs cross-node operation chain reconstruction. Taking a high-frequency abnormal communication event triggered by a user terminal as an example, the tracing process is as follows: Abnormal records marked as type C01 in the collaborative abnormal feature identifier set are located using transaction identifiers; based on the terminal device identifier carried by the record, the device operation logs generated by the associated edge nodes are retrieved; the session initiation records, target addresses, and network routing information in the logs are cross-analyzed; the security logs of the next-hop edge node are retrieved along the communication path, and the channel blocking operation status in the resource isolation policy execution records is verified; finally, a complete operation path graph spanning the terminal device, source edge node, relay node, and finally the target storage node is formed. During this process, the system automatically marks the role type, behavior time sequence, and resource interaction characteristics of the participating nodes in the operation chain.

[0048] Based on the reconstructed operation chain, the module generates a structured cross-domain operation tracing report. This report is divided into three main parts: the abnormal event sequence arranges all key operation events in a timeline, accurately marking the position and order of each event in the operation chain; the entity mapping diagram displays the relationships between participating nodes in a topological structure, using different graphic symbols to distinguish terminal, edge node, and storage node types, and marking the data transmission direction and control command flow on the connections; and the cross-node impact assessment section quantitatively analyzes the impact range of abnormal behavior, including assessment indicators such as the number of edge nodes affected by the resource isolation strategy, the proportion of total bandwidth resources occupied by the abnormal session, and the duration of interruptions to related tasks.

[0049] The strategy iteration update module receives the aforementioned source tracing report as input. This module first parses the cross-node impact assessment data. When multi-node resource overload is detected, it locates the communication pattern with abnormal bandwidth usage in the impact assessment. The module then triggers the multi-dimensional correlation analysis engine parameter update process of the collaborative feature analysis module, specifically adjusting the calculation model of the communication fluctuation index: extending the basic unit of the standard deviation statistics time window and increasing the duration requirement for abnormal fluctuation judgment. The parameter update command is transmitted in real-time to the operation queue of the collaborative feature analysis module through the system's internal interface, ensuring that the analysis engine uses the updated calculation standard to perform communication behavior analysis in the next processing cycle.

[0050] The data storage topology configuration output by the dynamic storage optimization module includes a list of storage node health statuses and a location mapping table. When this configuration changes, the edge task scheduling module automatically captures storage node status change events. If the status change involves the migration or expansion of core storage nodes, the edge task scheduling module initiates a refactoring process for the distributed computing task allocation strategy: recalculating the network transmission cost matrix based on the physical location of the new storage nodes; updating the edge node resource allocation ratio according to the health status list; and adjusting the preferred task routing path according to the location mapping table in the data storage topology configuration. The refactoring process maintains data synchronization with the real-time resource monitoring system to ensure that the new strategy conforms to the actual load capacity of each edge node.

[0051] The edge computing resource isolation policy generated by the security response control module is pushed to the collaborative behavior monitoring module in real time. The receiving end performs policy parsing and rule fusion: first, it extracts the restricted node list and resource access rules defined in the resource isolation policy; second, it matches and maps this list to the terminal registration database in the behavior monitoring; finally, it adjusts the preset collaborative mode parameters according to the security control level in the policy. A typical adjustment scenario is as follows: for terminal devices listed in the restricted node list, the collaborative behavior monitoring module lowers their document editing trajectory deviation tolerance threshold while increasing the sensitivity to abnormal communication session frequency. After rule fusion, the behavior monitoring module continuously monitors the changing trends of collaborative operation behavior flow under the new preset collaborative mode benchmark, forming a risk monitoring mechanism based on the latest security policy. Within this system, each module achieves dynamic policy transmission and execution status synchronization through standardized data interfaces, establishing a closed-loop governance architecture covering the entire process of anomaly detection, response control, and policy optimization.

[0052] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0053] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these 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. An enterprise-level intelligent office collaborative data processing system based on edge computing, characterized in that, The system includes: The collaborative data acquisition module is deployed on the enterprise's local edge node to capture collaborative operation behavior streams generated by multiple source office terminals in real time. The collaborative operation behavior streams include document editing trajectories, communication session frequencies, and task scheduling instruction sequences. The module extracts the spatiotemporal distribution features of the operation behaviors to generate original collaborative feature vectors. The collaborative feature analysis module receives the original collaborative feature vector, analyzes the coupling strength between the document editing trajectory and the task scheduling instruction sequence through a multi-dimensional correlation analysis engine, quantifies the fluctuation range of the communication session frequency within a preset time window, and integrates the coupling strength and fluctuation range to generate a quantified value of collaborative effectiveness. The edge task scheduling module dynamically divides the computing task load level based on the quantified value of the collaborative efficiency, matches the task execution priority according to the real-time resource utilization rate of the enterprise edge nodes, and generates a distributed computing task allocation strategy. The dynamic storage optimization module responds to the distributed computing task allocation strategy, monitors the data throughput latency of edge storage nodes, reconstructs the storage location index of frequently accessed data, and generates a data storage topology configuration across edge nodes.

2. The enterprise-level intelligent office collaborative data processing system according to claim 1, characterized in that, Also includes: The collaborative behavior monitoring module, based on the data storage topology configuration, tracks the trend of collaborative operation behavior flow changes of multi-source office terminals, identifies abnormal behavior characteristics that deviate from the preset collaborative mode, and generates a collaborative abnormal feature identifier set. The security response control module calls the collaborative abnormal feature identifier set, blocks the communication session channel associated with abnormal behavior features, resets the execution permission threshold of the task scheduling instruction sequence, and outputs the edge computing resource isolation strategy. The strategy iteration update module integrates the edge computing resource isolation strategy and the collaborative effectiveness quantification value, corrects the coupling strength calculation parameters of the multidimensional correlation analysis engine, and updates the spatiotemporal distribution feature extraction rules of the collaborative data acquisition module.

3. The enterprise-level intelligent office collaborative data processing system according to claim 1, characterized in that, The original collaborative feature vector includes the version iteration density of the document editing trajectory, the temporal distribution matrix of the communication session frequency, and the concurrent execution amount of the task scheduling instruction sequence; The quantifiable value of the synergistic effect includes the task coupling strength coefficient, communication fluctuation index, and resource matching score. The distributed computing task allocation strategy includes load balancing weights, task priority queues, and edge node resource allocation ratios. The data storage topology configuration defines the storage node location mapping table, data access path rules, and cross-node data synchronization frequency.

4. The enterprise-level intelligent office collaborative data processing system according to claim 1, characterized in that, The collaborative feature analysis module includes: The behavior coupling parsing submodule parses the dependency chain between the document editing trajectory and the task scheduling instruction sequence, calculates the overlapping execution frequency of key operation nodes in the dependency chain, and generates a task collaboration tightness index. The communication fluctuation monitoring submodule calculates the rate of change of the standard deviation of the communication session frequency within a continuous time unit, detects abnormal communication periods that exceed the preset fluctuation threshold, and generates a communication stability assessment matrix. The efficiency fusion calculation submodule aggregates the task collaboration tightness index and the communication stability evaluation matrix, generates a collaborative efficiency quantification value through weighted normalization processing, and outputs it to the edge task scheduling module.

5. The enterprise-level intelligent office collaborative data processing system according to claim 1, characterized in that, The edge task scheduling module includes: The load grading engine divides tasks into three levels—lightweight, medium-load, and high-load—based on the numerical range of collaborative performance quantification values, and correlates them with the real-time memory usage and CPU utilization of enterprise edge nodes. The priority matching unit calculates the preemption weight of different task levels on the enterprise edge node computing resources and generates a list of task execution priorities sorted by preemption weight; The allocation strategy generator dynamically adjusts the upper limit of task concurrency processing of enterprise edge nodes based on the task execution priority list, and generates a distributed computing task allocation strategy that includes node load allocation ratio and task queue scheduling sequence.

6. The enterprise-level intelligent office collaborative data processing system according to claim 1, characterized in that, The dynamic storage optimization module includes: The throughput monitoring unit continuously collects the data read / write response time and network transmission latency of edge storage nodes, and marks inefficient storage nodes whose response time exceeds a preset threshold. The index reconstruction engine analyzes the distribution density of frequently accessed data on inefficient storage nodes, migrates high-density data to nearby enterprise edge nodes, and rebuilds location index labels. The topology configuration generator generates a data storage topology configuration that includes cross-node data routing rules and a list of storage node health statuses, based on the reconstructed location index labels.

7. The enterprise-level intelligent office collaborative data processing system according to claim 2, characterized in that, The collaborative behavior monitoring module executes: Receive the list of storage node health statuses in the data storage topology configuration and monitor the collaborative operation flow of multi-source office terminals on healthy storage nodes; Compare the differences between the collaborative operation behavior flow and the preset collaborative mode in three dimensions: document editing trajectory deviation, communication session frequency deviation, and task scheduling sequence abnormal interruption rate. When the difference in any dimension exceeds the safety threshold, the abnormal behavior features are marked and a collaborative abnormal feature identifier set containing the abnormal type code and time stamp is generated.

8. The enterprise-level intelligent office collaborative data processing system according to claim 2, characterized in that, The security response control module includes: The session blocking unit parses the anomaly type code in the collaborative anomaly feature identifier set, freezes the transmission bandwidth of the corresponding communication session channel, and generates a channel isolation log. The permission reset engine reduces the maximum concurrent execution of associated task scheduling instruction sequences and sets new permission verification trigger conditions based on the severity level of the channel isolation logs. The resource isolation policy generator integrates channel isolation logs and permission verification trigger conditions to generate edge computing resource isolation policies that include a list of restricted nodes and a set of resource access rules.

9. The enterprise-level intelligent office collaborative data processing system according to claim 7, characterized in that, The strategy iteration update module executes: Extract the set of resource access rules from the edge computing resource isolation strategy and adjust the weight of the resource matching score calculation for the quantified value of collaborative effectiveness. Based on the adjusted resource matching score, the algorithm parameters of the task collaboration tightness index of the behavior coupling parsing submodule are corrected. The spatiotemporal distribution feature extraction rules of the collaborative data acquisition module are updated synchronously to increase the sampling density during periods of abnormal communication session frequency. The enterprise-level intelligent office collaborative data processing system according to claim 1, characterized in that it further includes: The cross-domain audit tracing module collects the operation records of the collaborative anomaly feature identifier set and the edge computing resource isolation strategy, and generates a cross-domain operation tracing report by associating the audit logs of multiple enterprise edge nodes; The cross-domain operation tracing report input strategy iterative update module triggers the multi-dimensional correlation analysis engine parameter update of the collaborative feature analysis module.

10. The enterprise-level intelligent office collaborative data processing system according to claim 2, characterized in that, The collaborative data acquisition module captures the collaborative operation behavior flow through lightweight probes deployed at edge nodes; The data storage topology configuration of the dynamic storage optimization module is fed back to the edge task scheduling module, triggering the reconstruction of the distributed computing task allocation strategy; The edge computing resource isolation strategy of the security response control module is synchronized to the collaborative behavior monitoring module, and the security threshold judgment benchmark of the preset collaborative mode is updated.

Images