A multi-live room linkage information monitoring processing method and system

Abstract

The application discloses a kind of multi-live room's linkage information monitoring processing method and system, it is related to the linkage monitoring technical field of multi-live room, including: unified identification and linkage relationship modeling to multiple live rooms, generate live room linkage topological structure;And the real-time data stream of each live room is multidimensional split, corresponding data sub-flow is generated;Lightweight processing pipeline is constructed to each data sub-flow, and stage-level asynchronous execution mechanism is introduced inside the pipeline, form asynchronous pending data state;According to asynchronous pending data state, real-time linkage event aggregation is carried out, and abnormal linkage mode is identified;According to abnormal linkage mode, corresponding risk level is generated, and the processing priority of data sub-flow is adjusted according to risk level;The application solves the problem that multi-live room concurrent data processing lags behind, real-time monitoring and intervention ability is insufficient.

Description

Technical Field

[0001] This invention relates to the field of multi-live-streaming room linkage monitoring technology, and more specifically, to a method and system for monitoring and processing linkage information of multiple live-streaming rooms. Background Technology

[0002] Multi-livestream room collaboration demands extremely high real-time performance, such as simultaneous broadcasting across multiple livestream rooms and second-level response to interactive commands. However, existing systems suffer from rigid computing power allocation strategies. Faced with massive amounts of concurrent data (comments, interactive behaviors, content tags) from multiple livestream rooms, traditional centralized data processing methods typically employ synchronous pipelines. This process suffers from severe serialization, easily leading to pipeline blockages and delayed monitoring results for collaborative events (e.g., violations are only identified minutes after they occur), missing the optimal intervention window. To address these issues, this invention proposes a solution. Summary of the Invention

[0003] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide a method and system for monitoring and processing information across multiple live streaming rooms. By constructing an asynchronous processing pipeline for multi-dimensional data sub-streams based on a linkage topology, and combining dynamic priority scheduling and risk level assessment methods, the method solves the problems of lagging concurrent data processing and insufficient real-time monitoring and intervention capabilities across multiple live streaming rooms.

[0004] To achieve the above objectives, the present invention provides the following technical solution:

[0005] Firstly, this application provides a method for monitoring and processing information related to the linkage of multiple live streaming rooms. The method includes: uniformly identifying and modeling the linkage relationships of multiple participating live streaming rooms to generate a linkage topology; based on the linkage topology, multi-dimensionally splitting the real-time data streams of each live streaming room and generating corresponding data sub-streams according to data type and linkage priority; constructing a lightweight processing pipeline for each data sub-stream and introducing a stage-level asynchronous execution mechanism within the pipeline, forming asynchronous pending data states through event-triggered data transmission and asynchronous buffering; aggregating real-time linkage events based on the asynchronous pending data states, and identifying abnormal linkage patterns based on the live streaming linkage topology and time window mechanism; generating corresponding risk levels based on the abnormal linkage patterns, and adjusting the processing priority of the data sub-streams according to the risk levels.

[0006] In one embodiment, multiple live streaming rooms participating in the linkage are uniformly identified and their linkage relationships are modeled to generate a live streaming room linkage topology. This includes: obtaining basic attribute information and generating a globally unified linkage identifier; collecting linkage trigger information based on the globally unified linkage identifier; abstracting the linkage trigger information into relationship edges and constructing a live streaming room linkage relationship graph model using the globally unified linkage identifier as nodes; calculating the relationship edge weights in the relationship graph model; classifying the live streaming room nodes into levels based on the relationship edge weights and the real-time online status of the live streaming room nodes; and generating a live streaming room linkage topology based on the level classification results and the live streaming room linkage relationship graph model.

[0007] In one embodiment, based on the live streaming room linkage topology, the real-time data stream of each live streaming room is split into multiple dimensions, and corresponding data sub-streams are generated according to data type and linkage priority. This includes: obtaining the linkage priority level of each live streaming room node in the live streaming room linkage topology; grouping the real-time data stream according to the live streaming room from which the data originates; splitting each group of live streaming room data streams into multiple dimensions based on data content characteristics to obtain multi-dimensional data sub-streams; and assigning priority labels to each data sub-stream based on the linkage priority level to generate a corresponding data sub-stream queue.

[0008] In one embodiment, generating the corresponding data sub-stream queue further includes: allocating processing units to each data sub-stream and collecting processing load information of each data sub-stream in real time; obtaining the processing delay status of the corresponding data sub-stream based on the processing load information; associating the processing load information, processing delay status, and linkage priority label of the data sub-stream to generate a computing power demand assessment result that characterizes the current urgency of the data sub-stream; and dynamically adjusting the computing power resource allocation ratio of the processing units corresponding to different data sub-streams based on the computing power demand assessment result.

[0009] In one embodiment, a lightweight processing pipeline is constructed for each data sub-stream, and a stage-level asynchronous execution mechanism is introduced within the pipeline. An asynchronous pending data state is formed through event-triggered data transfer and asynchronous buffering. This includes: constructing a lightweight processing pipeline for each data sub-stream and allocating an independent execution context; dividing the processing flow of the data sub-stream into multiple sequentially connected processing stages in the lightweight processing pipeline; and introducing an asynchronous execution mechanism in the lightweight processing pipeline so that adjacent processing stages can interact through asynchronous data transfer to form an asynchronous pending data state.

[0010] In one embodiment, an asynchronous execution mechanism is introduced into the lightweight processing pipeline, enabling adjacent processing stages to interact via asynchronous data transfer to form an asynchronous pending data state. This includes: defining an independent stage execution unit for each processing stage in the lightweight processing pipeline; constructing an event-triggered data transfer channel between the stage execution units corresponding to adjacent processing stages based on the stage execution units; configuring an asynchronous buffer structure for adjacent processing stages in the data transfer channel, and setting an asynchronous trigger condition corresponding to the asynchronous buffer structure for the subsequent processing stage; and enabling the stage execution units of multiple processing stages to run in parallel, thereby forming an asynchronous pending data state between adjacent processing stages.

[0011] In one embodiment, real-time linkage events are aggregated based on the status of asynchronous pending data, and abnormal linkage patterns are identified based on the live streaming room linkage topology and time window mechanism. This includes: classifying and associating asynchronous pending data statuses from different live streaming rooms according to the globally unified linkage identifier of the live streaming room to which each data substream belongs, forming a data set organized by live streaming room nodes; establishing a cross-live streaming room data dependency mapping for the data set based on the live streaming room linkage topology; applying a time window mechanism to the data results in the mapping, dividing the data into corresponding time periods according to their occurrence time; within each time window, aggregating and calculating data events from multiple live streaming rooms, including statistically analyzing the number of events, interaction frequency, content similarity, and behavioral sequence characteristics; and identifying abnormal linkage patterns based on the aggregation results, including cross-live streaming room abnormal interaction frequency, abnormal content duplication, and illegal redirection behavior.

[0012] In one embodiment, generating a corresponding risk level based on the abnormal linkage pattern includes: acquiring multi-dimensional indicator data of the abnormal linkage pattern within each time window and constructing an indicator matrix; normalizing the indicator matrix and calculating the weight ratio of each indicator; calculating the information entropy of the indicator based on the weight ratio; calculating the weight of the indicator based on the information entropy; calculating a comprehensive risk score based on the weight of each indicator and the normalized value; and classifying the abnormal events corresponding to the comprehensive risk score into risk levels according to a preset risk score threshold.

[0013] In one embodiment, adjusting the processing priority of data substreams based on risk level includes:

[0014] The priority coefficient of each data sub-stream is calculated based on the risk level and the original priority weight of the data sub-stream. Based on the priority coefficient, the queue of data sub-streams to be processed is sorted so that high-priority sub-streams are executed first in the processing pipeline. The adjusted data sub-stream order and priority coefficient are sent to the execution context of each lightweight processing pipeline. The lightweight processing pipeline dynamically adjusts the number of concurrent executions, resource allocation, and scheduling order according to the priority coefficient to achieve priority guarantee of computing power at the sub-stream level.

[0015] Secondly, this application provides a multi-live-stream room linkage information monitoring and processing system, which includes: a topology generation module, used to uniformly identify and model the linkage relationship of multiple live-stream rooms participating in the linkage, and generate a live-stream room linkage topology structure;

[0016] The data splitting module is used to split the real-time data stream of each live room in multiple dimensions based on the live room linkage topology, and generate corresponding data sub-streams according to data type and linkage priority.

[0017] The asynchronous processing module is used to build a lightweight processing pipeline for each data sub-stream and introduce a stage-level asynchronous execution mechanism inside the pipeline. It forms an asynchronous data state to be processed through event-triggered data transmission and asynchronous buffering.

[0018] The anomaly detection module is used to aggregate real-time linkage events based on the status of asynchronous pending data, and to identify abnormal linkage patterns based on the linkage topology and time window mechanism of the live broadcast room.

[0019] The intervention module is used to generate corresponding risk levels based on abnormal linkage patterns and adjust the processing priority of data sub-streams according to the risk levels.

[0020] As can be seen from the above technical solutions, the embodiments of this application have the following advantages:

[0021] 1. By uniformly identifying and modeling the participating live streaming rooms, a linkage topology is generated. The real-time data stream is then split into independent sub-streams according to the live streaming room, data type, and linkage priority, enabling asynchronous processing at the sub-stream level. Combined with a stage-level asynchronous execution mechanism, synchronous blocking between different sub-streams and processing stages is eliminated, significantly reducing overall processing latency. Simultaneously, an entropy weight method is used to calculate a comprehensive risk score for abnormal linkages and generate a risk level. Based on the risk level and the original priority of the sub-stream, the processing priority and computing power allocation are dynamically adjusted, enabling high-risk events to quickly obtain processing resources in a high-concurrency environment. This achieves real-time detection, rapid intervention, and efficient processing of abnormal linkages across live streaming rooms.

[0022] 2. Compared to conventional technical methods, this technical solution has several key advantages: First, it introduces a lightweight asynchronous pipeline at the sub-stream level and a stage-level asynchronous execution mechanism, decoupling processing and improving throughput and real-time performance in high-concurrency scenarios. Second, it combines the live-stream linkage topology and time window mechanism to aggregate real-time linkage events across live-stream rooms, enabling accurate identification of abnormal linkage patterns. Third, it dynamically generates indicator weights and risk scores based on the entropy weight method, solving the problem of poor adaptability of traditional experience-based weights or static threshold methods in high-concurrency, multi-live-stream scenarios. Fourth, it directly maps risk levels to sub-stream scheduling and computing power allocation, realizing a closed-loop dynamic scheduling mechanism for prioritizing abnormal events. This allows the system to respond to cross-live-stream abnormal behaviors within seconds, which is difficult to achieve with traditional single-threaded, static priority, or simple queue scheduling methods. Attached Figure Description

[0023] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0024] Figure 1 This is a schematic flowchart of a method for monitoring and processing information across multiple live streaming rooms, provided in an embodiment of this application.

[0025] Figure 2 This is a schematic diagram of the structure of a multi-live-streaming room linkage information monitoring and processing system provided in an embodiment of this application. Detailed Implementation

[0026] To enable those skilled in the art to better understand the technical solutions in this application, the technical solutions in the embodiments of this application will be clearly and completely described below. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0027] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this application, "a plurality of" or "several" means two or more, unless otherwise explicitly specified.

[0028] It should also be noted that, in this document, terms such as “comprising,” “including,” or any other variations thereof are intended to cover non-exclusive inclusion, such that an article or device that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such an article or device. Without further limitation, an element defined by the phrase “comprising one…” does not exclude the presence of other identical elements in the article or device that includes the aforementioned element.

[0029] Reference Figure 1 As shown in the diagram, the present invention provides a flowchart of a method for monitoring and processing information across multiple live streaming rooms, comprising the following steps:

[0030] S101 provides a unified identifier and models the linkage relationship among multiple live streaming rooms participating in the linkage, generating a corresponding live streaming room linkage topology to clarify the data association path and real-time priority between each live streaming room.

[0031] In this embodiment, multiple live streaming rooms participating in the linkage are uniformly identified and their linkage relationships are modeled to generate a corresponding live streaming room linkage topology, including:

[0032] The basic attribute information of multiple live rooms currently participating in the linkage is obtained from the live management system and data acquisition system. The basic attribute information includes the original unique identifier of the live room, the anchor account identifier, the broadcast timestamp, the channel identifier, and the status flag indicating whether it is currently in a linkage state.

[0033] Based on the aforementioned basic attribute information, a globally unified linkage identifier is generated for each live streaming room;

[0034] The globally unified linkage identifier is generated by combining the original unique identifier of the live room and the anchor account identifier, and a corresponding relationship is established with the original unique identifier through a mapping table, so as to uniquely identify the corresponding live room in data processing across systems and processing nodes.

[0035] Based on the globally unified linkage identifier, linkage trigger information between live streaming rooms is collected. The linkage trigger information includes synchronous broadcasting instructions, cross-live streaming room interactive behaviors, content sharing behaviors, and traffic guiding behaviors.

[0036] The linkage trigger information is subjected to relational abstraction processing, and the linkage relationship between live broadcast rooms is mapped as relation edges;

[0037] Using the globally unified linkage identifier of each live room as a node, and combining it with relation edges, a live room linkage relationship graph model is constructed, and the real-time online status of each live room and the number of historical linkages within a preset time window are recorded in the graph model.

[0038] Based on the historical number of linkages and the linkage type label corresponding to the relationship edge, the relationship edge weight is calculated for each relationship edge in the relationship graph model.

[0039] The edge weights are used to characterize the frequency and intensity of interaction between corresponding live streams within a preset time range. The edge weight calculation steps specifically include: within a preset statistical time window, counting the number of interaction trigger events between the two live streams associated with the corresponding edge to obtain the historical interaction count; then, based on the interaction type label corresponding to the edge, assigning different type influence coefficients to different types of interaction behaviors combined with historical experience, with interaction types having higher real-time requirements or greater propagation impact corresponding to higher influence coefficients; finally, normalizing and weighting the historical interaction counts and type influence coefficients to generate the edge weights, and recording them in the relationship graph model.

[0040] Based on the relationship edge weights and the real-time online status of the live room nodes, the live room nodes are hierarchically divided. The live rooms with high relationship edge weights and high activity are identified as core linkage nodes, and the remaining live rooms are identified as subordinate linkage nodes. The nodes are sorted in descending order according to the size of the relationship edge weights and the level of activity. A preset number of intervals are selected in order to generate linkage priority levels, including high level, medium level and low level.

[0041] Based on the hierarchical division results and the live room linkage relationship graph model, a live room linkage topology is generated. The linkage topology is used to describe the data propagation path between multiple live rooms, the diffusion direction of linkage events, and the real-time priority order of different nodes.

[0042] S102, based on the live streaming room linkage topology, the real-time data streams from each live streaming room are split into multiple dimensions. The real-time data streams include bullet screen data, interactive behavior data, and content tag data, and corresponding data sub-streams are generated according to data type and linkage priority.

[0043] In this embodiment, based on the live streaming room linkage topology, the real-time data streams from each live streaming room are split into multiple dimensions, and corresponding data sub-streams are generated according to data type and linkage priority, including:

[0044] Obtain the linkage priority level of each live room node in the live room linkage topology;

[0045] The real-time data streams from each live streaming room are grouped according to the live streaming room from which the data originates, and the real-time data streams are bound to the globally unified linkage identifier of the corresponding live streaming room.

[0046] Based on the data content characteristics, each group of live room data streams is split into multiple dimensions, including: bullet screen data sub-stream: containing all text, emoticons and gift information sent by users; interactive behavior data sub-stream: containing records of interactive behaviors such as likes, votes, Q&A, and lotteries; and content tag data sub-stream: containing tags, classification information and key themes generated from the live room video content.

[0047] Based on the linkage priority level of the live room nodes, the split data sub-streams are prioritized. The data sub-streams of high-priority live rooms are marked for priority processing to ensure that the data of key linkage nodes are processed and transmitted faster. The priority marking includes three levels: high, medium, and low.

[0048] Generate corresponding data sub-stream queues. Each data sub-stream queue is bound to its respective live streaming room and data type. The data sub-stream queue includes the correspondence between the queue identifier and the globally unified linkage identifier of the live streaming room, the data type identifier (bullet comments, interactive behavior or content tags), priority label and timestamp information.

[0049] The data substream queues are then added to the pipeline to be processed in order of priority.

[0050] It should be noted that the approach of splitting the real-time data stream into corresponding sub-streams based on the live-stream linkage topology has significant technical advantages: by splitting the original mixed data stream into independent sub-streams according to the live-stream room, data type, and linkage priority, it can not only achieve parallel and asynchronous processing of bullet comments, interactive behaviors, and content tags, avoiding processing blockage between different types of data, but also use priority labeling to ensure that the data of key linkage nodes are processed first, thereby improving the response speed and processing efficiency of cross-live-stream linkage events; at the same time, the attached queue structure ensures that the order of data is controllable and traceable in high-concurrency scenarios, providing accurate and implementable basic data support for subsequent dynamic computing power scheduling, abnormal linkage identification, and real-time intervention, significantly reducing processing latency and enhancing the real-time performance and scalability of the system.

[0051] Furthermore, generating the corresponding data substream queue also includes:

[0052] At least one processing unit is allocated to each data substream, the processing unit being a computing instance for performing data substream processing tasks, and a load and latency monitoring module is enabled in the processing unit;

[0053] The load and latency monitoring module collects the processing load information of each data sub-stream in real time based on the queue status information and processing count information inside the processing unit. The processing load information includes the current amount of data to be processed, the length of the processing queue, and the number of data processed per unit time.

[0054] Based on the processing load information, the actual processing time of each data sub-stream is calculated using the timestamps recorded when the data enters the processing queue and when the processing is completed. The actual processing time is then compared with a preset delay threshold to obtain the processing delay status of the corresponding data sub-stream.

[0055] The processing load information, processing delay status, and data sub-stream linkage priority label are associated to generate a computing power demand assessment result that characterizes the current urgency of the data sub-stream.

[0056] Based on the computing power demand assessment results, the computing power resource allocation ratio of the processing units corresponding to different data sub-streams is dynamically adjusted. Specifically, for data sub-streams with high linkage priority and processing delay exceeding a preset delay threshold, the available computing power resource ratio of their processing units is increased, so that the computing power resources that the processing unit can obtain per unit time are higher than those of the processing units corresponding to low linkage priority data sub-streams.

[0057] After adjusting the computing power resource allocation ratio, the resource usage status information of each processing unit is updated, and the updated status information is fed back to the load and latency monitoring module to form a dynamic computing power scheduling closed loop based on real-time monitoring results. Through this dynamic computing power scheduling closed loop, high-priority data sub-streams continuously receive higher computing power resource support than low-priority data sub-streams in high-concurrency processing scenarios, thereby reducing the processing latency of critical linked data and improving the real-time response capability of the overall system. Data sub-streams refer to the original real-time data streams from various live streaming rooms, divided into independent processing units according to live streaming room, data type (bullet comments, interactive behavior, content tags), and linkage priority, for parallel processing, scheduling, and computing power allocation in a lightweight asynchronous pipeline.

[0058] S103 constructs a lightweight processing pipeline for each data sub-stream and introduces a stage-level asynchronous execution mechanism within the pipeline, forming an asynchronous pending data state through event-triggered data transfer and asynchronous buffering.

[0059] In this embodiment, a lightweight processing pipeline is constructed for each data sub-stream, and a stage-level asynchronous execution mechanism is introduced within the pipeline. This mechanism uses event-triggered data transfer and asynchronous buffering to form an asynchronous data state to be processed, including:

[0060] A corresponding lightweight processing pipeline is constructed for each data sub-stream. The lightweight processing pipeline is an independent processing flow for processing a single data sub-stream, and is used to execute the parsing, processing and output tasks of the corresponding data sub-stream.

[0061] Each lightweight processing pipeline is assigned an independent execution context, which includes processing status information, task scheduling information, and resource usage identifiers to isolate the processing status and resource usage of different data sub-streams, so as to avoid resource competition and status interference between different data sub-streams.

[0062] In each lightweight processing pipeline, the processing flow of the data substream is divided into multiple sequentially connected processing stages, which include a data parsing stage, a business logic processing stage, and a result output stage.

[0063] An asynchronous execution mechanism is introduced into the lightweight processing pipeline, enabling adjacent processing stages to interact through asynchronous data transfer. After the previous processing stage completes data delivery, it can continue to execute the next processing task without waiting for the next processing stage to complete its processing, thus forming an asynchronous pending data state between adjacent processing stages.

[0064] It's worth noting that by constructing an independent lightweight processing pipeline for each data sub-stream and introducing a stage-level asynchronous execution mechanism within the pipeline, each processing stage can interact through event-triggered data transfer and asynchronous buffering. This avoids synchronous blocking caused by the previous stage waiting for the next stage to complete, achieving sub-stream-level processing decoupling and improving high-concurrency processing capabilities. This design not only reduces the overall data processing chain's latency, improving system throughput and real-time performance, but also ensures that the processing status of each data sub-stream is independent and can be perceived in real time. This provides a reliable, low-latency raw data foundation for subsequent cross-live-stream aggregation analysis based on the asynchronous pending data status, and for quickly identifying abnormal linkage patterns or synchronous behaviors, enabling abnormal behaviors to be detected and handled promptly.

[0065] Furthermore, an asynchronous execution mechanism is introduced into the lightweight processing pipeline, enabling adjacent processing stages to interact via asynchronous data transfer, forming an asynchronous pending data state, including:

[0066] Each processing stage in the lightweight processing pipeline is defined with an independent stage execution unit. Each stage execution unit is used to encapsulate the processing logic of the corresponding processing stage and is equipped with a data receiving interface for receiving input data and a result output interface for outputting processing results.

[0067] Based on the aforementioned stage execution unit, an event-triggered data transmission channel is constructed between the stage execution units corresponding to adjacent processing stages. The data transmission channel is used to carry the data results output by the previous processing stage and sends a processing trigger signal to the next processing stage in the form of an event when the data arrives.

[0068] An asynchronous buffer structure is configured for adjacent processing stages in the data transmission channel. The asynchronous buffer structure is used to temporarily store the data results output by the previous processing stage, so that the previous processing stage can immediately release execution resources after completing the data writing, without waiting for the next processing stage to complete the data processing.

[0069] An asynchronous trigger condition corresponding to the asynchronous buffer structure is set for the next processing stage. The asynchronous trigger condition is generated based on the data arrival event or buffer state change event in the asynchronous buffer structure, and is used to drive the next processing stage to actively pull and process the corresponding data when available computing power resources are available.

[0070] In the lightweight processing pipeline, the stage execution units of multiple processing stages are run in parallel, so that the previous processing stage does not need to wait synchronously with the next processing stage after completing the data delivery, thereby forming an asynchronous pending data state between adjacent processing stages.

[0071] The asynchronous pending data state refers to an intermediate state in which the data is waiting to be retrieved and processed by the subsequent processing stage when the previous processing stage has completed the delivery of the processing result and released the execution resources, but the next processing stage has not yet started or completed the processing of the data.

[0072] It should be noted that this technical solution does not simply use thread pools or message queues to achieve asynchronous processing. Instead, it breaks down the processing logic into independently schedulable stage execution units within the data sub-stream level processing pipeline and introduces a data transfer mechanism based on event triggering and asynchronous buffering. This makes asynchronous execution an inherent property of the pipeline structure, eliminating synchronous blocking between adjacent processing stages at the architectural level.

[0073] S104 aggregates real-time linkage events based on the status of asynchronous pending data, and identifies abnormal linkage patterns based on the linkage topology and time window mechanism of the live broadcast room.

[0074] In this embodiment, real-time linkage events are aggregated based on the status of asynchronous pending data, and abnormal linkage patterns are identified based on the live broadcast room linkage topology and time window mechanism, including:

[0075] Based on the globally unified linkage identifier of the live room to which each data substream belongs, the asynchronous pending data status from different live rooms is classified and associated to form a data set organized by live room node;

[0076] The asynchronous pending data status includes data event type, occurrence timestamp, source live stream identifier, user identifier, and interaction behavior type, ensuring that each piece of data can be accurately attributed to the live stream and event category.

[0077] Based on the live streaming room linkage topology, a cross-live streaming room data dependency mapping is established for the classified and associated data sets. The mapping is used to describe the possible synchronization behaviors and linkage paths between different live streaming rooms, providing a structural basis for subsequent aggregation analysis.

[0078] A time window mechanism is applied to the data results in the mapping, dividing the data into corresponding time periods according to their occurrence time, so as to analyze cross-live room synchronization behavior within the same time window.

[0079] Within each time window, data events from multiple live streaming rooms are aggregated and calculated, including the number of events, interaction frequency, content similarity, and behavioral sequence characteristics.

[0080] The content similarity can be calculated based on text matching or content tag similarity algorithms, and the behavioral sequence features include the order of interactive behaviors such as liking, voting, Q&A, or lottery, forming a cross-live room synchronous behavioral indicator.

[0081] Based on the aggregation results, abnormal linkage patterns are identified, including abnormal interaction frequency across live streaming rooms, abnormal content duplication, and illegal traffic redirection behavior.

[0082] The identification of abnormal linkage patterns based on aggregation results includes the following steps: First, for the synchronous behavior indicators across live streaming rooms within each time window, the interaction frequency, user interaction repetition rate, content tag similarity, and number of illegal redirection behaviors between each live streaming room are extracted; then, these indicators are compared with preset abnormal judgment thresholds; if the interaction frequency between some live streaming rooms is higher than the normal range, user interaction behaviors or content are highly repetitive (i.e., the user interaction repetition rate and content tag similarity exceed the preset threshold), or the number of illegal redirection behaviors exceeds the preset standard number, it is judged as an abnormal linkage pattern; finally, abnormal labeling information is generated for each identified abnormality, including the live streaming room ID involved, abnormal type, time window, and abnormal indicator value, so that the system can provide real-time alarms, interventions, or trigger policy execution for abnormal linkages, thereby achieving timely detection and response to abnormal behaviors across live streaming rooms.

[0083] It's important to note that real-time aggregation analysis using asynchronous pending data status allows for rapid statistical analysis of interactive behaviors across live streams, content similarity calculations, and behavioral sequence analysis. This enables timely identification of abnormal cross-live stream interaction patterns, including abnormal interaction frequency, content duplication, and illegal redirection behaviors. The advantages are twofold: firstly, introducing a stage-level asynchronous execution mechanism within the pipeline eliminates blocking between different data substreams and processing stages, improving overall system throughput and real-time responsiveness; secondly, low-latency aggregation-based anomaly identification enables rapid detection and intervention of cross-live stream interaction behaviors, allowing operators to take measures in the shortest possible time to prevent the spread of violations or impact on user experience, while ensuring the system's scalability and reliability in high-concurrency, multi-live stream scenarios.

[0084] S105 generates a corresponding risk level based on the abnormal linkage mode, and adjusts the processing priority of subsequent data sub-streams according to the risk level to achieve continuous and rapid response to high-risk linkage behaviors.

[0085] In this embodiment, a corresponding risk level is generated based on the abnormal linkage mode, including:

[0086] Obtain multi-dimensional indicator data of abnormal linkage patterns within each time window, and construct an indicator matrix. ,in This represents the value of the i-th abnormal event on the j-th indicator. The multidimensional indicator data includes the interaction frequency between each live broadcast room, the user interaction repetition rate, the content tag similarity, and the number of illegal redirection behaviors.

[0087] The index matrix is ​​normalized.

[0088]

[0089] In the formula, Let j be the normalized value of the i-th anomalous event on the j-th indicator. Let j be the minimum value of the j-th indicator in the current dataset. It represents the maximum value of the j-th indicator in the current dataset.

[0090] After normalization, the weight ratio of each indicator is calculated.

[0091]

[0092] In the formula, Let m be the percentage of the i-th abnormal event in the j-th indicator, and m be the number of abnormal events.

[0093] Calculate the information entropy of the j-th indicator based on the weight ratio;

[0094]

[0095] In the formula, Let k be the information entropy of the j-th indicator. This is the normalization coefficient.

[0096] Calculate the weight of the j-th indicator based on the entropy value;

[0097]

[0098] In the formula, Let be the weight of the j-th indicator, and n be the total number of indicators.

[0099] The weights of each indicator are compared with the normalized values. The comprehensive risk score is obtained by multiplying the results and then summing them by weight.

[0100] Based on the preset risk scoring threshold, abnormal events corresponding to the comprehensive risk score are classified into high, medium, and low risk levels.

[0101] Specifically, if the risk scoring threshold is less than the first risk scoring threshold, it is a low-risk level; if the risk scoring threshold exceeds the first risk scoring threshold but is less than the second risk scoring threshold, it is a medium-risk level; and if the risk scoring threshold exceeds the second risk scoring threshold, it is a high-risk level.

[0102] It should be noted that by normalizing each abnormal indicator and calculating the information entropy based on the distribution of the indicators within the current time window, the weight of each indicator is automatically generated. This does not rely on historical label data and can adapt to changes in cross-live room data in real time. At the same time, it ensures that indicators of different dimensions and magnitudes can be fairly compared and weighted summed. Furthermore, by highlighting key indicators with large fluctuations and high information content, the final risk score more accurately reflects the real abnormal linkage behavior and improves the system's ability to quickly identify abnormal events in high-concurrency, multi-live room scenarios.

[0103] Furthermore, the processing priority of subsequent data substreams is adjusted according to the risk level, including:

[0104] Based on the risk level and the original priority weights of the data sub-streams, i.e., the relation edge weights, the priority coefficient of each data sub-stream is calculated through a mapping function, specifically:

[0105]

[0106] In the formula, This is the priority coefficient. For the preset risk level weights, The original priority weights of the data substreams, i.e., the relation edge weights, , These are adjustable coefficients, obtained through historical experience. This is a priority mapping function that maps the risk score of abnormal events and the original priority of data substreams to a unified priority coefficient.

[0107] The data sub-streams to be processed are sorted according to priority coefficients so that high-priority sub-streams are executed first in the processing pipeline. The sorting rules include sorting in descending order of priority factor.

[0108] The adjusted data substream order and priority coefficients are distributed to the execution context of each lightweight processing pipeline.

[0109] The lightweight processing pipeline dynamically adjusts the number of concurrent executions, resource allocation, and scheduling order based on priority coefficients to ensure priority computing power at the sub-stream level. When the number of high-priority sub-streams increases or the processing delay exceeds the threshold, the system increases the number of concurrent threads in the corresponding pipeline or allocates more computing resources to ensure that high-risk events are processed in the shortest possible time.

[0110] It should be noted that by combining the risk level of abnormal events with the original priority weights of data substreams, a unified priority coefficient is calculated, and data substreams are dynamically sorted and scheduled accordingly. This ensures that data from high-risk events can receive priority computing resources and processing time in the lightweight processing pipeline. Combined with a real-time adjustment mechanism for concurrency and resource allocation, priority computing power is guaranteed at the substream level. This enables rapid response to abnormal linkage behaviors in high-concurrency scenarios with multiple live streaming rooms, significantly reducing processing latency and improving the overall system throughput and real-time intervention efficiency.

[0111] Reference Figure 2 As shown in the diagram, the present invention provides a schematic diagram of a multi-live-stream room linkage information monitoring and processing system, which includes a topology generation module, a data splitting module, an asynchronous processing module, an anomaly identification module, and an intervention module. The modules are interconnected.

[0112] The topology generation module is used to uniformly identify and model the linkage relationship of multiple live rooms participating in the linkage, and generate the linkage topology structure of the live room.

[0113] The data splitting module is used to split the real-time data stream of each live room in multiple dimensions based on the live room linkage topology, and generate corresponding data sub-streams according to data type and linkage priority.

[0114] The asynchronous processing module is used to build a lightweight processing pipeline for each data sub-stream and introduce a stage-level asynchronous execution mechanism inside the pipeline. It forms an asynchronous data state to be processed through event-triggered data transmission and asynchronous buffering.

[0115] The anomaly detection module is used to aggregate real-time linkage events based on the status of asynchronous pending data, and to identify abnormal linkage patterns based on the linkage topology and time window mechanism of the live broadcast room.

[0116] The intervention module is used to generate corresponding risk levels based on abnormal linkage patterns and adjust the processing priority of data sub-streams according to the risk levels.

[0117] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, in the form of a computer program product.

[0118] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0119] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0120] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0121] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for monitoring and processing information across multiple live streaming rooms, characterized in that, Includes the following steps: A unified identifier and linkage relationship model are created for multiple live streaming rooms participating in the collaboration, generating a live streaming room linkage topology. Based on the live streaming room linkage topology, the real-time data stream of each live streaming room is split into multiple dimensions, and corresponding data sub-streams are generated according to data type and linkage priority. A lightweight processing pipeline is constructed for each data sub-stream, and a stage-level asynchronous execution mechanism is introduced inside the pipeline. The asynchronous pending data state is formed through event-triggered data transmission and asynchronous buffering. The asynchronous pending data state includes data event type, occurrence timestamp, source live room identifier, user identifier, and interaction behavior type. Real-time linkage events are aggregated based on the status of asynchronous pending data, and abnormal linkage patterns are identified based on the live broadcast room linkage topology and time window mechanism, including: Based on the globally unified linkage identifier of the live room to which each data substream belongs, the asynchronous pending data status from different live rooms is classified and associated to form a data set organized by live room node; Based on the interconnected topology of live streaming rooms, a cross-live streaming room data dependency mapping is established for the data set; A time window mechanism is applied to the data results in the mapping, dividing the data into corresponding time periods according to their occurrence time. Within each time window, data events from multiple live streaming rooms are aggregated and calculated, including the number of events, interaction frequency, content similarity, and behavioral sequence characteristics. Based on the aggregation results, abnormal linkage patterns are identified, including abnormal interaction frequency across live streaming rooms, abnormal content duplication, and illegal traffic redirection behavior. The corresponding risk level is generated based on the abnormal linkage mode, and the processing priority of the data sub-stream is adjusted according to the risk level.

2. The method for monitoring and processing information across multiple live streaming rooms according to claim 1, characterized in that, The process of unifying the identification and modeling the linkage relationships of multiple participating live streaming rooms to generate a live streaming room linkage topology includes: Obtain basic attribute information and generate a globally unified linkage identifier; Based on a globally unified linkage identifier, linkage trigger information is collected; The linkage trigger information is abstracted into relation edges, and a globally unified linkage identifier is used as a node to construct a live broadcast room linkage relation graph model. Calculate the weights of the relation edges in the relation graph model; Based on the relation edge weights and the real-time online status of the live streaming room nodes, the live streaming room nodes are hierarchically divided. Based on the hierarchical division results and the live room linkage relationship graph model, a live room linkage topology structure is generated.

3. The method for monitoring and processing information across multiple live streaming rooms according to claim 1, characterized in that, The method, based on the live streaming room linkage topology, involves multi-dimensionally splitting the real-time data streams of each live streaming room and generating corresponding data sub-streams according to data type and linkage priority, including: Obtain the linkage priority level of each live room node in the live room linkage topology; Group the real-time data stream according to the live broadcast room from which the data originates. Based on the data content characteristics, each group of live room data streams is split into multi-dimensional sub-streams to obtain multi-dimensional data sub-streams; Based on the linkage priority level, each data sub-stream is marked with priority, and a corresponding data sub-stream queue is generated.

4. The method for monitoring and processing information across multiple live streaming rooms according to claim 3, characterized in that, The generation of the corresponding data sub-stream queue also includes: Each data substream is assigned a processing unit, and the processing load information of each data substream is collected in real time. The processing delay status of the corresponding data sub-stream is obtained based on the processing load information; By associating processing load information, processing latency status, and the linkage priority label of data sub-streams, a computing power demand assessment result is generated to characterize the current urgency of the data sub-streams. Based on the results of the computing power demand assessment, the allocation ratio of computing power resources for processing units corresponding to different data sub-streams is dynamically adjusted.

5. The method for monitoring and processing information across multiple live streaming rooms according to claim 1, characterized in that, The process involves constructing a lightweight processing pipeline for each data sub-stream and introducing a stage-level asynchronous execution mechanism within the pipeline. This mechanism uses event-triggered data transfer and asynchronous buffering to create asynchronous data states awaiting processing, including: Build lightweight processing pipelines for each data substream and allocate independent execution contexts; In a lightweight processing pipeline, the processing flow of data substreams is divided into multiple sequentially connected processing stages; Introducing an asynchronous execution mechanism into a lightweight processing pipeline enables adjacent processing stages to interact through asynchronous data transfer, forming an asynchronous pending data state.

6. The method for monitoring and processing information across multiple live streaming rooms according to claim 5, characterized in that, The introduction of an asynchronous execution mechanism into the lightweight processing pipeline enables interaction between adjacent processing stages through asynchronous data transfer, forming an asynchronous pending data state, including: Define an independent stage execution unit for each processing stage in the lightweight processing pipeline; Based on the aforementioned stage execution unit, an event-triggered data transmission channel is constructed between the stage execution units corresponding to adjacent processing stages; In the data transmission channel, an asynchronous buffer structure is configured for adjacent processing stages, and an asynchronous trigger condition corresponding to the asynchronous buffer structure is set for the next processing stage. This allows the execution units of multiple processing stages to run in parallel, thereby creating asynchronous data states to be processed between adjacent processing stages.

7. The method for monitoring and processing information across multiple live streaming rooms according to claim 1, characterized in that, The generation of corresponding risk levels based on abnormal linkage patterns includes: Obtain multi-dimensional indicator data of abnormal linkage patterns within each time window and construct an indicator matrix; The indicator matrix is ​​normalized, and the weight ratio of each indicator is calculated. The information entropy of the index is calculated based on the proportion, and the weight of the index is calculated based on the information entropy; The comprehensive risk score is calculated based on the weights of each indicator and the normalized values. Based on the preset risk scoring threshold, the abnormal events corresponding to the comprehensive risk score are classified into risk levels.

8. The method for monitoring and processing information across multiple live streaming rooms according to claim 1, characterized in that, The adjustment of data substream processing priority based on risk level includes: The priority coefficient of each data sub-stream is calculated based on the risk level and the original priority weight of the data sub-stream; Based on priority coefficients, the queue of data sub-streams to be processed is sorted so that high-priority sub-streams are executed first in the processing pipeline; The adjusted data substream order and priority coefficients are distributed to the execution context of each lightweight processing pipeline. The lightweight processing pipeline dynamically adjusts the number of concurrent executions, resource allocation, and scheduling order based on priority coefficients to ensure priority computing power at the sub-stream level.

9. A system using the multi-live-stream room linkage information monitoring and processing method as described in any one of claims 1-8, characterized in that, include: The topology generation module is used to uniformly identify and model the linkage relationship of multiple live rooms participating in the linkage, and generate the linkage topology structure of the live room. The data splitting module is used to split the real-time data stream of each live room in multiple dimensions based on the live room linkage topology, and generate corresponding data sub-streams according to data type and linkage priority. The asynchronous processing module is used to build a lightweight processing pipeline for each data sub-stream and introduce a stage-level asynchronous execution mechanism inside the pipeline. It forms an asynchronous data state to be processed through event-triggered data transmission and asynchronous buffering. The anomaly detection module is used to aggregate real-time linkage events based on the status of asynchronous pending data, and to identify abnormal linkage patterns based on the linkage topology and time window mechanism of the live broadcast room. The intervention module is used to generate corresponding risk levels based on abnormal linkage patterns and adjust the processing priority of data sub-streams according to the risk levels.

Images