Method and system for screening and matching candidate sets of working subgraphs of computing resource graphs

Abstract

The present invention provides a method and system for screening and matching candidate sets of working subgraphs of a computing resource graph, including: candidate set screening adopts a two-level candidate set mode, and by constructing a two-level screening mechanism, a subgraph candidate set that meets the conditions is screened out from the computing resource graph in real time; the corresponding matching method uses a specific query algorithm based on the node connection relationship, resource configuration parameters, performance indicators and other information stored in the candidate set to accurately find all subgraphs that can execute computing tasks in the best way. The present invention can effectively avoid the waste and imbalance of computing resources in big data distributed clusters while ensuring the optimal computing speed. In the big data distributed resource computing scenario, the candidate set screening and matching method for working subgraphs of the computing resource graph proposed by the present invention shows good screening and matching efficiency.

Description

Technical Field

[0001] The present invention relates to the field of big data computing technology, specifically, to a method and system for screening and matching working subgraph candidate sets of computing resource graphs, and more specifically, to a method for screening and matching working subgraph candidate sets of computing resource graphs in a big data distributed dynamic resource scenario. Background Art

[0002] With the rapid development of information technology, big data has been widely used in various fields. In big data distributed resource computing scenarios, computing resource graphs, as an important tool for describing the computing resource structure and the connections between resources, play a key role in the effective allocation of resources, the rational scheduling of tasks, and the optimization of system performance.

[0003] A big data distributed cluster is a system composed of multiple computing devices connected via a network, capable of collaboratively handling large-scale data storage, computation, and analysis tasks. Its core features include distributed storage, parallel computing, elastic scalability, and high availability. In such scenarios, the proper allocation of computing resources directly determines the efficiency and cost of task execution. Improper resource allocation leads to load imbalance, which can significantly increase task completion time. Real-time data processing tasks are sensitive to network latency. Insufficient link bandwidth between computing nodes can cause significant data processing delays and even system crashes.

[0004] A computing resource graph intuitively displays the organization and interaction paths of various resources within a computing cluster. Nodes in the graph represent computing resources, such as servers, storage devices, and GPU clusters, while edges represent data transmission links, dependencies, or collaborative working relationships between resources. By analyzing the computing resource graph, administrators can clearly understand the resource distribution of the entire computing cluster, including information such as resource type, performance parameters, and load status, providing an important basis for resource planning, allocation, and scheduling. In big data distributed resource computing, different business needs and data processing characteristics place diverse requirements on the combination and connection of computing resources. For example, deep learning training tasks require a large number of GPU resources and high-bandwidth data transmission links to ensure efficient model training; real-time streaming data processing tasks require low-latency computing resources and fast data exchange channels to ensure timely data processing.

[0005] Currently, the screening and matching of candidate work subgraphs is a key research direction in the study and application of computing resource graphs. A work subgraph refers to the portion of a computing resource graph related to a specific task or business. It contains the various computing resources and their connections required to complete the task. In practical applications, it is often necessary to screen candidate work subgraphs that meet specific criteria from complex computing resource graphs and match them with actual business needs to achieve precise resource allocation and efficient task execution. However, existing methods for screening and matching candidate work subgraphs have several limitations. On the one hand, with the continuous expansion of computing clusters and the increasing complexity of data processing tasks, traditional screening and matching algorithms have high computational complexity and low efficiency when processing large-scale computing resource graphs, making it difficult to meet real-time requirements. For example, in ultra-large-scale cloud computing centers, computing resource graphs may contain hundreds of thousands of nodes and links. Using traditional algorithms to screen and match work subgraphs can be time-consuming and computationally resource-intensive. On the other hand, existing methods lack the ability to account for the dynamic changes in computing resources. The resource environment in big data distributed resource computing scenarios is dynamically changing, with factors such as computing node failures and repairs, resource elastic scaling, and real-time adjustments to business needs. Traditional methods often struggle to quickly adapt to these changes, resulting in reduced accuracy and effectiveness in screening and matching results. Furthermore, different business scenarios have diverse requirements for working subgraphs, and existing general-purpose methods struggle to meet the personalized needs of various complex business scenarios. For example, in high-frequency trading systems in the financial sector, the stability, low latency, and security of computing resources are extremely demanding, requiring the selection of working subgraphs with high reliability and strong security protection attributes. In meteorological big data simulation scenarios, however, greater emphasis is placed on the large-scale parallel processing capabilities of computing resources, and the screening and matching criteria for working subgraphs differ from those in the financial sector.

[0006] For the candidate set screening and matching of working subgraphs of computing resource graphs in big data distributed resource computing scenarios, a more efficient, flexible and dynamically adaptable method is needed to improve the utilization of computing resources and the efficiency of business processing and meet the diverse needs of different business scenarios.

[0007] Patent document CN117648466A (application number: 202311661711.9) discloses a distributed subgraph matching method and device. Through the master node, all strongly connected components in the pattern graph are regarded as a pattern subgraph, which reduces the computational difficulty and resource consumption of complex pattern graph matching. The master node obtains calculation information based on the pattern subgraph and distributes it to all nodes in sequence, so that all nodes search for matching subgraphs that match the pattern subgraph in the target graph, and then obtain a matching result set and upload it to a distributed storage system. The master node distributes the matching result set to the slave nodes according to the connection relationship between the pattern subgraphs to merge the matching result sets, obtain the target result set and return it to the master node for storage in the distributed storage system. By implementing distributed parallel processing, the computational speed and horizontal scalability of graph matching are improved, and it can cope with ultra-large-scale data graphs and pattern graphs. Summary of the Invention

[0008] In view of the defects in the prior art, the purpose of the present invention is to provide a method and system for screening and matching working subgraph candidate sets of a computing resource graph.

[0009] According to the present invention, a method for screening and matching a candidate set of a working subgraph of a computing resource graph is provided, comprising:

[0010] Step S1: Analyze the connection relationship between nodes in the computing resource graph D, preliminarily screen the nodes based on preset criteria, and generate a global candidate set GC;

[0011] Step S2: updating edge connection information according to dynamic changes of network nodes in the computing resource graph D, updating the global candidate set GC based on the updated edge connection information, further screening the updated global candidate set GC based on preset criteria to generate a temporary candidate set LC; wherein the temporary candidate set LC includes an index structure;

[0012] Step S3: Obtain a candidate working subgraph set based on the working subgraph Q and the temporary candidate set LC, and select the optimal working subgraph based on the candidate working subgraph set according to the load balancing strategy.

[0013] Preferably, step S1 includes: constructing a screening model based on quantitative indicators including neighbor label frequency; based on node attributes, connection relationships and computing task requirements, using the screening model to preliminarily screen redundant nodes that are irrelevant to the task objectives to obtain a global candidate set GC.

[0014] Preferably, the step S1 includes:

[0015] Assign attribute labels to each node in the computing resource graph D to identify the resource type and performance parameter characteristics of the node; for any node , count the label distribution of its neighbor nodes and calculate the neighbor label frequency; suppose the label of node u is A, if it has k neighbors with label B, then the screening model is: the neighbor label frequency of label B of candidate point u is greater than or equal to k;

[0016] For a node with label A , use the screening model to filter nodes that meet If there are k or more neighbors with label B, then the node Incorporate Node The global candidate points of

[0017] Traverse all nodes in the working subgraph Q, repeatedly trigger the above steps, and collect all nodes that meet the conditions into a global candidate set GC.

[0018] Preferably, step S2 includes: updating node-associated edge information in real time based on network dynamic change events, updating the global candidate set based on the updated node-associated edge information, performing secondary screening on the updated global candidate set using a screening model, and establishing an index structure including node topology features, resource status, and task matching degree to form a temporary candidate set LC.

[0019] Preferably, the step S2 includes: monitoring the computing resource graph D in real time, and triggering an update of the temporary candidate set when a node joins or disconnects.

[0020] Preferably, step S3 includes:

[0021] The matching order of the working subgraph Q is determined according to the number of candidate points for each node in the working subgraph Q in the temporary candidate set LC, and a set of candidate working subgraphs is obtained according to the matching order; the optimal working subgraph is selected based on the candidate working subgraph set according to the load balancing strategy.

[0022] According to the present invention, a system for screening and matching candidate sets of working subgraphs of a computing resource graph is provided, comprising:

[0023] Module M1: Analyzes the connection relationship between nodes in the computing resource graph, preliminarily screens the nodes based on preset criteria, and generates a global candidate set of GCs;

[0024] Module M2: updates edge connection information based on dynamic changes in network nodes in the computing resource graph, updates the global candidate set GC based on the updated edge connection information, further screens the updated global candidate set GC based on preset criteria, and generates a temporary candidate set LC; wherein the temporary candidate set LC includes an index structure;

[0025] Module M3: Obtain a candidate working subgraph set based on the working subgraph Q and the temporary candidate set LC, and select the optimal working subgraph based on the candidate working subgraph set according to the load balancing strategy.

[0026] Preferably, the module M1 includes: constructing a screening model based on quantitative indicators including neighbor label frequency; using the screening model to preliminarily screen redundant nodes irrelevant to the task target based on node attributes, connection relationships and computing task requirements to obtain a global candidate set GC;

[0027] The module M1 includes:

[0028] Assign attribute labels to each node in the computing resource graph D to identify the resource type and performance parameter characteristics of the node; for any node , count the label distribution of its neighbor nodes and calculate the neighbor label frequency; suppose the label of node u is A, if it has k neighbors with label B, then the screening model is: the neighbor label frequency of label B of candidate point u is greater than or equal to k;

[0029] For a node with label A , use the screening model to filter nodes that meet If there are k or more neighbors with label B, then the node Incorporate Node The global candidate points of

[0030] Traverse all nodes in the working subgraph Q, repeatedly trigger the above steps, and collect all nodes that meet the conditions into a global candidate set GC.

[0031] Preferably, the module M2 includes: updating node-related edge information in real time based on network dynamic change events, updating a global candidate set based on the updated node-related edge information, performing secondary screening on the updated global candidate set using a screening model, and establishing an index structure including node topology features, resource status, and task matching degree to form a temporary candidate set LC;

[0032] The module M2 includes: real-time monitoring of the computing resource graph D, and triggering the update of the temporary candidate set when a node joins or disconnects.

[0033] Preferably, the module M3 includes:

[0034] The matching order of the working subgraph Q is determined according to the number of candidate points for each node in the working subgraph Q in the temporary candidate set LC, and a set of candidate working subgraphs is obtained according to the matching order; the optimal working subgraph is selected based on the candidate working subgraph set according to the load balancing strategy.

[0035] Compared with the prior art, the present invention has the following beneficial effects:

[0036] 1. The present invention can effectively avoid the waste and imbalance of computing resources while ensuring the optimal computing speed;

[0037] 2. In the large-scale distributed resource computing scenario, the candidate set screening and matching method of the computing resource graph working subgraph proposed in this invention shows good matching efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Other features, objects and advantages of the present invention will become more apparent upon reading the detailed description of non-limiting embodiments with reference to the following drawings:

[0039] Figure 1 Flowchart of the method for screening and matching candidate sets of working subgraphs for computing resource graphs.

[0040] Figure 2a This is a diagram obtained after resource abstraction of a big data distributed system.

[0041] Figure 2b A graph of the resources required for calculation.

[0042] Figure 3 For Figure 2a Schematic diagram after deleting an edge from the foundation.

[0043] Figure 4 For Figure 3 Schematic diagram after adding an edge to the foundation.

[0044] Figure 5 Schematic diagram of the changes in the global candidate set GC.

[0045] Among them, N stands for NPU, C stands for CPU, and G stands for GPU. DETAILED DESCRIPTION

[0046] The present invention will be described in detail below with reference to specific embodiments. The following examples will help those skilled in the art to further understand the present invention, but are not intended to limit the present invention in any form. It should be noted that, for those skilled in the art, several changes and improvements can be made without departing from the scope of the present invention. These all fall within the scope of protection of the present invention.

[0047] Example 1

[0048] According to the present invention, a method for screening and matching candidate sets of working subgraphs of a computing resource graph is provided, such as Figure 1 Shown, including:

[0049] Step S1: Analyze the connection relationship between nodes in the computing resource graph D, preliminarily screen the nodes based on preset criteria, and generate a global candidate set GC;

[0050] Step S2: updating edge connection information according to dynamic changes of network nodes in the computing resource graph D, updating the global candidate set GC based on the updated edge connection information, further screening the updated global candidate set GC based on preset criteria to generate a temporary candidate set LC; wherein the temporary candidate set LC includes an index structure;

[0051] Step S3: Obtain a candidate working subgraph set based on the working subgraph Q and the temporary candidate set LC, and select the optimal working subgraph based on the candidate working subgraph set according to the load balancing strategy.

[0052] Specifically, step S1 includes: constructing a screening model based on quantitative indicators including neighbor label frequency; based on node attributes, connection relationships and computing task requirements, using the screening model to preliminarily screen redundant nodes that are irrelevant to the task objectives to obtain a global candidate set GC.

[0053] Specifically, step S1 includes:

[0054] The computing resource graph includes:

[0055] ;

[0056] Among them, D represents the computing resource graph; V represents the node set, which is the computing resource entity. , each node Represents a physical or virtual computing resource, including servers, GPUs, storage devices, etc.; E represents an edge set, which is a communication link between nodes. , each edge express and There is a direct communication path between them; L represents the label function, which assigns one or more labels to each node to describe the resource type or applicable task type; T represents the time function, which records the state update time of the node or edge to support dynamic change perception.

[0057] Assign attribute labels to each node in the computing resource graph D to identify the resource type and performance parameter characteristics of the node; for any node , count the label distribution of its neighbor nodes and calculate the neighbor label frequency; suppose the label of node u is A, if it has k neighbors with label B, then the screening model is: the neighbor label frequency of label B of candidate point u is greater than or equal to k;

[0058] For a node with label A , use the screening model to filter nodes that meet If there are k or more neighbors with label B, then the node Incorporate Node The global candidate points of

[0059] Traverse all nodes in the working subgraph Q, repeatedly trigger the above steps, and collect all nodes that meet the conditions into a global candidate set GC.

[0060] Specifically, step S2 includes: updating node-associated edge information in real time based on network dynamic change events, updating the global candidate set based on the updated node-associated edge information, performing secondary screening on the updated global candidate set using the screening model, and establishing an index structure including node topology features, resource status, and task matching degree to form a temporary candidate set LC.

[0061] Specifically, step S2 includes: monitoring the computing resource graph in real time, and triggering the update of the temporary candidate set when a node joins or disconnects.

[0062] Specifically, step S3 includes:

[0063] The matching order of the working subgraph Q is determined according to the number of candidate points for each node in the working subgraph Q in the temporary candidate set LC, and a set of candidate working subgraphs is obtained according to the matching order; the optimal working subgraph is selected based on the candidate working subgraph set according to the load balancing strategy.

[0064] Example 2

[0065] According to the present invention, a method and system for screening and matching a candidate set of a working subgraph of a computing resource graph are provided, which realizes accurate scheduling of computing resources and rapid execution of tasks through a secondary screening mechanism and an efficient matching strategy.

[0066] Among them, in the screening of secondary candidate sets, a hierarchical filtering architecture is adopted: the first-level screening is based on node attributes, connection relationships and computing task requirements, and a screening model is constructed through quantitative indicators such as neighbor label frequency to quickly eliminate redundant nodes irrelevant to the task goals and obtain a global candidate set; the second-level screening is based on network dynamic change events, including: node joining and link disconnection, real-time updating of node-related edge information, and further updating of the global candidate set. The updated global candidate set is deep pruned in combination with the screening model, and an index structure containing node topology characteristics, resource status and task matching is established to form a streamlined temporary candidate set, which greatly compresses the invalid search space.

[0067] The matching phase employs an efficient, index-driven graph matching strategy. This includes parsing the topological structure features, resource configuration parameters, and computational task constraints stored in the candidate set index to construct a priority matching model. This prioritizes subgraphs with high correlation and low load for matching, enabling rapid location and selection of working subgraphs. Finally, based on a subgraph load balancing evaluation model, the optimal working subgraph is dynamically selected to execute the computational task, effectively improving task scheduling efficiency and resource utilization in distributed resource computing scenarios.

[0068] The present invention also provides a working subgraph candidate set screening and matching system for a computing resource graph. The working subgraph candidate set screening and matching system for a computing resource graph can be implemented by executing the process steps of the working subgraph candidate set screening and matching method for the computing resource graph. That is, those skilled in the art can understand the working subgraph candidate set screening and matching method for the computing resource graph as an optimal implementation method of the working subgraph candidate set screening and matching system for the computing resource graph.

[0069] Example 3

[0070] Example 3 is a preferred example of Example 1

[0071] According to the present invention, a method for screening and matching a candidate set of a working subgraph of a computing resource graph is provided, comprising:

[0072] The method is applied to a big data distributed resource computing environment, involving a fully distributed resource network topology, a resource subgraph (working subgraph) Q required for the current computation, a global candidate set GC, a temporary candidate set LC, a set M of all currently matching working subgraphs, and a set of matching working subgraphs that disappear / appear after edges are updated. Δ M. Through the secondary candidate set screening mechanism and efficient matching algorithm, it can achieve rapid screening and accurate matching of working subgraphs in the network topology graph;

[0073] The specific steps include:

[0074] Global candidate set construction steps: Based on the neighbor label frequency (NLF) metric, perform coarse-grained screening of the network topology graph to generate a global candidate set. GC;

[0075] Temporary candidate set update steps: In response to dynamic changes in network nodes, including: node joining / disconnecting, updating edge connection information, updating the global candidate set GC based on the updated edge connection information, performing secondary pruning based on the updated global candidate set, and generating a temporary candidate set LC .

[0076] Working subgraph matching step: Using the temporary candidate set index information, the optimized graph matching algorithm is used to quickly locate the subgraph set that needs to be updated. ΔM , and update the candidate working subgraph set M, and then select the optimal subgraph on the candidate working subgraph set M based on the load balancing strategy to execute the computing task.

[0077] Specifically, the global candidate set construction step includes:

[0078] Define the node label and neighbor label frequency sub-step: assign attribute labels to each node in the network topology to identify the node's resource type, performance parameters and other characteristics. , count the label distribution of its neighboring nodes and calculate the neighbor label frequency (NLF). Suppose the label of node u is A. If it has k neighbors with label B, then the NLF of label B of u's candidate point must be greater than or equal to k.

[0079] Screening global candidate points sub-step: For the node with label A , if it meets the following conditions, it will be included in the global candidate points of node u: node v has k or more neighbors with label B (where k is the NLF of the B label of node u.

[0080] Generate global candidate set sub-step: traverse all nodes in the working subgraph Q, repeat the above steps, and form the global candidate set GC with all nodes that meet the conditions.

[0081] Specifically, the temporary candidate set updating step includes:

[0082] Monitor the network topology in real time. When a node joins ins(v1,v2) or disconnects del(v1,v2) events, the temporary candidate set update process is triggered.

[0083] Update edge connection information based on dynamic changes of network nodes in the computing resource graph;

[0084] The global candidate set GC is updated based on the updated edge connection information. The updated global candidate set is screened again using the same screening method, and an index structure including node topology features, resource status, and task matching degree is established to form a temporary candidate set LC.

[0085] The working subgraph matching step includes:

[0086] For the temporary candidate set LC that has been obtained, the matching order of the working subgraph is determined according to the number of candidate points in the temporary candidate set LC, and the set of changed working subgraphs is obtained according to the matching order. Δ M, then make M Δ The change of M results in a new M. More specifically, if an edge is added, it should be M←M∪ΔM, and if an edge is deleted, it should be M←M-ΔM;

[0087] Finally, when the computation is to be performed, the most appropriate work subgraph is selected from M according to the load balancing strategy.

[0088] Example 4

[0089] Example 4 is a preferred example of Example 1

[0090] Ultra-large-scale big data computing clusters typically deploy hundreds of thousands of heterogeneous computing resources. These resources exhibit significant heterogeneity: CPUs are suitable for general-purpose computing tasks; GPUs excel at parallel computing and vector operations; and NPUs are designed specifically for neural network acceleration. Furthermore, hardware resources are synergistically adapted to each other. For example, high-performance CPUs require large-capacity memory to maximize their advantages in compute-intensive tasks, while low-speed CPUs combined with high-capacity storage are more suitable for I / O-intensive tasks. Furthermore, the connectivity status (bandwidth, latency, and reliability) of each resource node in the network topology also impacts task execution efficiency.

[0091] The above computing resources are abstracted as nodes in the distributed computing resource topology graph. Node labels identify the task type for which the resource is applicable (e.g., CPU for general computing, GPU for deep learning, NPU for model reasoning). Edges between nodes represent direct communication links between resources. The global resource graph is maintained by the global management node in a master-slave architecture, where:

[0092] Master node: responsible for resource screening, task allocation, and real-time update of the global resource map;

[0093] Slave node: realizes hot backup through data synchronization mechanism, and undertakes redundant computing and load balancing tasks to ensure high availability of the system.

[0094] When a big data computing task like Spark or Flink is submitted to a management node, the system generates a corresponding computational graph based on the task's characteristics, known as a work subgraph Q. This generation process involves analyzing key task parameters, such as data scale (input / output data volume); data storage location (data warehouse, data lake); computation type (batch processing, real-time stream computing, machine learning training, etc.); and performance requirements (concurrency, response time requirements, and data transmission bandwidth requirements).

[0095] After the work subgraph Q is generated, the system performs the following resource screening and matching process:

[0096] 1. Global candidate set construction: Based on the Neighbor Label Frequency (NLF) algorithm, the initial candidate node set GC that matches Q is screened in the global resource graph;

[0097] 2. Work subgraph matching: Further screening from the GC generates a set M of candidate work subgraphs that meet the task requirements;

[0098] 3. Load balancing scheduling: Through the load evaluation model (taking into account indicators such as CPU load, memory usage, and network latency), the optimal working subgraph is selected from M to execute the task.

[0099] To cope with dynamic changes in resources, the management node monitors network topology changes (such as node failures and link interruptions) and resource status fluctuations (such as resource expansion and task load changes) in real time. When the global resource map is updated:

[0100] 1. Candidate set synchronization update: triggering incremental update of the global candidate set GC;

[0101] 2. Temporary candidate set calculation: Combine the updated GC and change information to generate a temporary candidate set LC;

[0102] 3. Dynamic adjustment of working subgraphs: The changed working subgraph set ΔM is calculated through LC and merged with the existing candidate working subgraph set M to update.

[0103] When the same or similar tasks are subsequently submitted, the system can quickly select the appropriate work subgraph from the updated M based on the load balancing strategy, significantly reducing resource screening time. Through the above mechanism, the present invention achieves dynamic perception, efficient screening, and intelligent scheduling of resources in large-scale computing clusters, significantly improving resource utilization and task execution efficiency in big data computing.

[0104] Example 5

[0105] Figure 2a This is a distribution diagram of big data computing resources, where C, N, and G represent CPU, NPU, and GPU, respectively. They are abstracted as labels in the diagram, and the edges between the points represent that two computing resources can communicate directly. Figure 2b is the working subgraph required for a computing task. Figure 2a to Figure 2b You can see: yes Candidate nodes; yes Candidate nodes; yes Candidate nodes; yes Candidate nodes; that is:

[0106] .

[0107] Therefore, at this time, if you want to perform calculations, you can select the subgraph {v1, v3, v5, v7} for calculation.

[0108] Figure 3 This means that the edge (v5, v7) has been deleted, which means that the two computing resources v5 and v7 can no longer communicate directly. The entire communication link will change. The changed communication link is:

[0109] ;

[0110] If the edge is deleted, it should be M←M-ΔM; therefore, after deleting the edge, M is updated to an empty set, and there are no matching resources in the graph to satisfy the execution of this calculation.

[0111] Figure 4 This means that (v2, v7) is reconnected. At this time, the communication link changes:

[0112] .

[0113] If the edge is added, it should be M←M∪ΔM. The original M is an empty set, so the current M is updated to:

[0114] ;

[0115] Among them, M represents the candidate working subgraph set, represents the set of candidate working subgraphs to be updated, Represents a node on the computing resource graph, Represents a node on the working subgraph.

[0116] At this time, the computing task can select the subgraph {v2, v3, v6, v7} for calculation.

[0117] During this period, the changes in the global candidate set are as follows Figure 5 shown.

[0118] Those skilled in the art will appreciate that, in addition to implementing the system, device, and various modules provided by the present invention in purely computer-readable program code, it is entirely possible to implement the same program in the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers, embedded microcontrollers, and the like by logically programming the method steps. Therefore, the system, device, and various modules provided by the present invention can be considered a hardware component, and the modules included therein for implementing various programs can also be considered structures within the hardware component; the modules for implementing various functions can also be considered both software programs for implementing the method and structures within the hardware component.

[0119] The above describes specific embodiments of the present invention. It should be understood that the present invention is not limited to the specific embodiments described above, and those skilled in the art may make various changes or modifications within the scope of the claims, which do not affect the essence of the present invention. The embodiments of this application and the features in the embodiments may be combined with each other in any manner unless there is a conflict.

Claims

1. A method for screening and matching candidate sets of working subgraphs of a computing resource graph, characterized in that: include: Step S1: Analyze the connection relationship between nodes in the computing resource graph D, preliminarily screen the nodes based on preset criteria, and generate a global candidate set GC; Step S2: updating edge connection information according to dynamic changes of network nodes in the computing resource graph D, updating the global candidate set GC based on the updated edge connection information, further screening the updated global candidate set GC based on preset criteria to generate a temporary candidate set LC; wherein the temporary candidate set LC includes an index structure; Step S3: Obtain a candidate working subgraph set based on the working subgraph Q and the temporary candidate set LC, and select the optimal working subgraph based on the candidate working subgraph set according to the load balancing strategy; The method for screening and matching working subgraph candidate sets of computing resource graphs is applied to a big data distributed resource computing environment, and realizes rapid screening and accurate matching of working subgraphs in a network topology graph through a secondary candidate set screening mechanism and matching algorithm. The step S1 includes: constructing a screening model based on quantitative indicators including neighbor label frequencies; using the screening model to preliminarily screen redundant nodes that are irrelevant to the task objectives based on node attributes, connection relationships, and computing task requirements, to obtain a global candidate set GC; The step S1 comprises: Assign attribute labels to each node in the computing resource graph D to identify the resource type and performance parameter characteristics of the node; for any node , count the label distribution of its neighbor nodes and calculate the neighbor label frequency; suppose the label of node u is A, if it has k neighbors with label B, then the screening model is: the neighbor label frequency of label B of candidate point u is greater than or equal to k; For a node with label A , use the screening model to filter nodes that meet If there are k or more neighbors with label B, then the node Incorporate Node The global candidate points of Traverse all nodes in the working subgraph Q, repeatedly trigger the above steps, and combine all nodes that meet the conditions into a global candidate set GC; The step S2 includes: updating node-related edge information in real time based on network dynamic change events, updating a global candidate set based on the updated node-related edge information, performing a secondary screening on the updated global candidate set using a screening model, and establishing an index structure including node topology features, resource status, and task matching degree to form a temporary candidate set LC; The step S2 includes: monitoring the computing resource graph in real time, and triggering the update of the temporary candidate set when a node joins or disconnects.

2. The method for screening and matching a candidate set of a computing resource graph working subgraph according to claim 1, characterized in that: The step S3 comprises: The matching order of the working subgraph Q is determined according to the number of candidate points for each node in the working subgraph Q in the temporary candidate set LC, and a set of candidate working subgraphs is obtained according to the matching order; the optimal working subgraph is selected based on the candidate working subgraph set according to the load balancing strategy.

3. A system for screening and matching candidate sets of working subgraphs of a computing resource graph, characterized in that: include: Module M1: Analyzes the connection relationship between nodes in the computing resource graph, preliminarily screens the nodes based on preset criteria, and generates a global candidate set of GCs; Module M2: updates edge connection information based on dynamic changes in network nodes in the computing resource graph, updates the global candidate set GC based on the updated edge connection information, further screens the updated global candidate set GC based on preset criteria, and generates a temporary candidate set LC; wherein the temporary candidate set LC includes an index structure; Module M3: Obtain a candidate working subgraph set based on the working subgraph Q and the temporary candidate set LC, and select the optimal working subgraph based on the candidate working subgraph set according to the load balancing strategy; The working subgraph candidate set screening and matching system of the computing resource graph is applied to the big data distributed resource computing environment, and realizes the rapid screening and accurate matching of working subgraphs in the network topology graph through the secondary candidate set screening mechanism and matching algorithm; The module M1 includes: building a screening model based on quantitative indicators including neighbor label frequency; using the screening model to preliminarily screen redundant nodes that are irrelevant to the task target based on node attributes, connection relationships and computing task requirements, to obtain a global candidate set GC; The module M1 includes: Assign attribute labels to each node in the computing resource graph D to identify the resource type and performance parameter characteristics of the node; for any node , count the label distribution of its neighbor nodes and calculate the neighbor label frequency; suppose the label of node u is A, if it has k neighbors with label B, then the screening model is: the neighbor label frequency of label B of candidate point u is greater than or equal to k; For a node with label A , use the screening model to filter nodes that meet If there are k or more neighbors with label B, then the node Incorporate Node The global candidate points of Traverse all nodes in the working subgraph Q, repeatedly trigger the above steps, and combine all nodes that meet the conditions into a global candidate set GC; The module M2 includes: updating node-related edge information in real time based on network dynamic change events, updating the global candidate set based on the updated node-related edge information, performing secondary screening on the updated global candidate set using a screening model, and establishing an index structure including node topology features, resource status, and task matching degree to form a temporary candidate set LC; The module M2 includes: real-time monitoring of the computing resource graph, and triggering the update of the temporary candidate set when a node joins or disconnects.

4. The system for screening and matching candidate sets of working subgraphs of a computing resource graph according to claim 3, characterized in that: The module M3 includes: The matching order of the working subgraph Q is determined according to the number of candidate points for each node in the working subgraph Q in the temporary candidate set LC, and a set of candidate working subgraphs is obtained according to the matching order; the optimal working subgraph is selected based on the candidate working subgraph set according to the load balancing strategy.

Images