Routing method and device based on intention-aware path, computer equipment, readable storage medium and program product

Abstract

The invention relates to a routing method and device based on an intention-aware path, computer equipment, a computer readable storage medium and a computer program product. By adopting the method, the user behavior data can be processed through the multi-dimensional intention recognition model, the target user intention of the user can be quickly and accurately obtained, screening is carried out in the computing power routing network, the intention sensing path is generated, low-delay guiding of service routing is achieved based on the intention sensing path, and the user experience is improved. According to the invention, direct perception of the computing power demand and direct analysis of the high-level demand of the user are realized, and the resource utilization rate, the security and the performance of the computing power scheduling system are improved, so that the user demand can be responded more quickly and accurately.

Description

Technical Field

[0001] The present application relates to the field of network communication technology, and in particular to a routing method, apparatus, computer equipment, computer-readable storage medium, and computer program product based on intent-aware paths. Background Art

[0002] With the rapid development of computing power networks, the problem of uneven distribution of computing resources between nodes has become increasingly prominent. In order to better schedule computing power equipment from a global perspective, it is necessary to schedule computing power across domains. In traditional technologies, computing power resources are generally allocated and scheduled based on traditional routing protocols. However, since the above traditional routing protocols do not have the ability to directly perceive computing power demand, the scheduling efficiency of computing power resources is low. Summary of the invention

[0003] Based on this, it is necessary to provide a routing method, device, computer equipment, computer-readable storage medium and computer program product based on intent-aware paths that can meet the computing power routing requirements with high efficiency and low latency in response to the above technical problems.

[0004] In a first aspect, the present application provides a routing method based on an intent-aware path, comprising:

[0005] Acquiring user behavior data, wherein the user behavior data includes access requirements and / or resource operation behaviors;

[0006] Processing the user behavior data through a multi-dimensional intention recognition model to obtain a plurality of first encoding results corresponding to the user behavior data;

[0007] For each node in the computing power routing network, determine an initial candidate node for the next node of the node; and based on the node routing table of the node and the first encoding result, screen each of the initial candidate nodes to obtain a target candidate node;

[0008] Based on each of the target candidate nodes, an intent-aware path is generated, and computing resources corresponding to the intent-aware path are determined in a computing resource pool.

[0009] In one embodiment, the determining of the initial candidate node of the next node of the node; and screening each of the initial candidate nodes based on the node routing table of the node and the first encoding result to obtain the target candidate node, includes:

[0010] Based on the topological structure of the computing power routing network, nodes that are directly connected to the node are determined as initial candidate nodes for the next node of the node;

[0011] In each of the initial candidate nodes, based on the node routing table of the node, determine a second encoding result of the user intention satisfied by each of the initial candidate nodes;

[0012] Based on the second encoding results of the user intentions satisfied by each of the initial candidate nodes and the multiple first encoding results corresponding to the user behavior data, a target candidate node is obtained by screening.

[0013] In one embodiment, the second encoding results based on the user intentions satisfied by each of the initial candidate nodes and the plurality of first encoding results corresponding to the user behavior data are screened to obtain the target candidate node, including:

[0014] Based on the multiple first encoding results corresponding to the user behavior data, screening among the multiple initial candidate nodes based on the second encoding results respectively satisfied by the initial candidate nodes, to obtain first candidate nodes matching the first encoding results;

[0015] Based on a preset mapping relationship between encoding results and user intents, determine the target user intents corresponding to the first encoding results, and calculate the cost value of each of the first candidate nodes by weighting the target user intents and the index value corresponding to each of the target user intents in the first candidate nodes;

[0016] Screening is performed based on the cost value of each of the first candidate nodes to obtain a target candidate node.

[0017] In one of the embodiments, the intent-aware path includes multiple paths; and determining, in a computing resource pool, computing resources corresponding to the intent-aware path includes:

[0018] In the computing resource pool including multiple types of computing resources, respectively determine the computing resources corresponding to the destination network endpoints of each of the paths;

[0019] Based on the priority of each type of computing power resources, the computing power resources corresponding to each of the destination network endpoints are screened to obtain the computing power resources corresponding to the intent-aware path.

[0020] In one embodiment, the method further comprises:

[0021] User feedback data is obtained, and the mapping relationship is updated based on the user feedback data to obtain an updated mapping relationship, each of the mapping relationships being configured with version information.

[0022] In one embodiment, the encoding result is a color encoding result; the mapping relationship between the preset encoding result and the user intention is a mapping relationship between the preset color encoding result and the user intention; the method further includes:

[0023] Collecting user evaluation information on the target user's intention, and determining a color coding result that meets a preset obfuscation condition based on the evaluation information;

[0024] Based on the color coding results that meet the preset confusion conditions, the mapping relationship is improved to obtain an improved mapping relationship.

[0025] In a second aspect, the present application also provides a routing device based on an intent-aware path, comprising:

[0026] A first acquisition module, used to acquire user behavior data, wherein the user behavior data includes access requirements and / or resource operation behaviors;

[0027] A first determination module, configured to process the user behavior data through a multi-dimensional intention recognition model to obtain a plurality of first encoding results corresponding to the user behavior data;

[0028] A second determination module is used to determine, for each node in the computing power routing network, an initial candidate node of the next node of the node; and to screen each of the initial candidate nodes based on the node routing table of the node and the first encoding result to obtain a target candidate node;

[0029] The third determination module is used to generate an intention-aware path based on each of the target candidate nodes, and determine the computing power resources corresponding to the intention-aware path in the computing power resource pool.

[0030] In a third aspect, the present application further provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:

[0031] Acquire user behavior data, wherein the user behavior data includes access requirements and / or resource operation behaviors; process the user behavior data through a multi-dimensional intent recognition model to obtain a plurality of first encoding results corresponding to the user behavior data; determine, for each node in the computing power routing network, an initial candidate node for the next node of the node; and based on the node routing table of the node and the first encoding result, screen each of the initial candidate nodes to obtain a target candidate node; generate an intent-aware path based on each of the target candidate nodes, and determine the computing power resources corresponding to the intent-aware path in the computing power resource pool.

[0032] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the following steps are implemented:

[0033] Acquire user behavior data, wherein the user behavior data includes access requirements and / or resource operation behaviors; process the user behavior data through a multi-dimensional intent recognition model to obtain a plurality of first encoding results corresponding to the user behavior data; determine, for each node in the computing power routing network, an initial candidate node for the next node of the node; and based on the node routing table of the node and the first encoding result, screen each of the initial candidate nodes to obtain a target candidate node; generate an intent-aware path based on each of the target candidate nodes, and determine the computing power resources corresponding to the intent-aware path in the computing power resource pool.

[0034] In a fifth aspect, the present application further provides a computer program product, including a computer program, which implements the following steps when executed by a processor:

[0035] Acquire user behavior data, wherein the user behavior data includes access requirements and / or resource operation behaviors; process the user behavior data through a multi-dimensional intent recognition model to obtain a plurality of first encoding results corresponding to the user behavior data; determine, for each node in the computing power routing network, an initial candidate node for the next node of the node; and based on the node routing table of the node and the first encoding result, screen each of the initial candidate nodes to obtain a target candidate node; generate an intent-aware path based on each of the target candidate nodes, and determine the computing power resources corresponding to the intent-aware path in the computing power resource pool.

[0036] The above-mentioned routing method, device, computer equipment, computer-readable storage medium and computer program product based on intent-aware path, by adopting this method, processes user behavior data through a multi-dimensional intent recognition model, quickly and accurately obtains the user's target user intention, and screens it in the computing power routing network to generate an intent-aware path, and realizes low-latency guidance of business routing based on the intent-aware path, realizes direct perception of computing power requirements and direct analysis of users' high-level needs, improves the resource utilization, security and performance of the computing power scheduling system, and enables user needs to be responded to more quickly and accurately. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the drawings required for use in the embodiments of the present application or related technical descriptions will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0038] Figure 1is an application environment diagram of a routing method based on intent-aware paths in one embodiment;

[0039] Figure 2 A schematic diagram of a flow chart of a routing method based on an intent-aware path in an embodiment;

[0040] Figure 3 A schematic diagram of a process for determining a target candidate node in an embodiment;

[0041] Figure 4 A schematic diagram of a flow chart of a step of determining a target candidate node in another embodiment;

[0042] Figure 5 A schematic diagram of a flow chart of steps for allocating computing resources in one embodiment;

[0043] Figure 6 A schematic diagram of a process for improving a mapping relationship in an embodiment;

[0044] Figure 7 is a schematic diagram of a communication architecture in an embodiment;

[0045] Figure 8 A schematic diagram of the structure of a computing power routing node in one embodiment;

[0046] Fig. 9 is a flowchart of a routing method based on intent-aware paths in another embodiment;

[0047] Fig.10 A schematic diagram of a process of transmitting a data stream in an embodiment;

[0048] Fig.11 is a structural block diagram of a routing device based on intent-aware paths in one embodiment;

[0049] Fig.12 FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0050] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0051] The routing method based on the intention-aware path provided in the embodiment of the present application can be applied to Figure 1In the application environment of the system shown, the system includes a user 102, a network device 104, a network 106, a computing power routing network 108, and a computing power resource pool 110. The network device 104 can communicate with each computing power routing node included in the computing power routing network 108 through the network 106, and then allocate corresponding computing power resources to the user 102 in the computing power resource pool 110.

[0052] In an exemplary embodiment, a routing method based on intent-aware paths is provided, and the method is applied to a system as an example for explanation. The system can be a server. Optionally, the server can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides cloud computing services. The system includes at least a network device 104 and a computing power routing network 108. The network device 104 can include a user identity authentication module, a routing module, a computing power service routing encapsulation module, and an intent analysis module. The computing power routing network 108 can include multiple computing power routing nodes, and each computing power routing node can include a computing resource perception module, a routing decision module, a BGP extension module, and a computing power service routing identification module. Figure 2 As shown, the method in this embodiment may include:

[0053] Step 202, obtaining user behavior data.

[0054] Among them, user behavior data includes access requirements and / or resource operation behaviors.

[0055] Specifically, the user can first perform identity authentication services, for example, the identity of the user accessing the system can be verified through the user identity authentication module. After the verification is passed, the user can access and operate in the system. The system can record the computing power network resource requirements and operation behavior of the user's access, that is, obtain the user's operation behavior data. The system can store the above user's operation behavior data in the form of system logs or event tracking. The identity authentication service in the user identity authentication module can be OAuth, LDAP or SAML.

[0056] Step 204: Process the user behavior data through a multi-dimensional intention recognition model to obtain multiple first encoding results corresponding to the user behavior data.

[0057] Among them, the multi-dimensional intent recognition model can be a multi-dimensional data model based on user demand characteristics, which is used to map and analyze different user intentions, such as low-latency transmission, low cost, etc. The multi-dimensional deep model can construct multiple data dimensions through a multi-dimensional data cube, determine each data dimension as a type of data feature, and query and filter data through multiple data dimensions. Correspondingly, the multi-dimensional intent recognition model can be a deep learning model or a machine learning model, such as a long short-term memory network, a BERT model, etc.

[0058] Specifically, the system can input user behavior data into the multi-dimensional intent recognition model, analyze and recognize the user intent in the user behavior data through the multi-dimensional intent recognition model, and obtain the output result of the multi-dimensional intent recognition model, which can be multiple encoding results (hereinafter referred to as the first encoding result), that is, the system can input user behavior data into the multi-dimensional intent recognition model, and transform the user behavior data through the multi-dimensional intent recognition model to obtain multiple encoding results. The encoding result can be a color encoding result.

[0059] Step 206, for each node in the computing power routing network, determine the initial candidate node of the next node of the node; and based on the node routing table of the node and the first encoding result, screen the initial candidate nodes to obtain the target candidate node.

[0060] Among them, the computing power routing network may include multiple computing power routing nodes, and each computing power routing node is connected through a network topology; each computing power routing node may include a computing resource perception module, a routing decision module, a BGP extension module, and a computing power service routing identification module.

[0061] Specifically, after obtaining one or more first coding results corresponding to the user behavior data, the system can obtain multiple target candidate nodes that can meet the target user's intention based on each first coding result. In one example, for each computing power routing node that receives the first coding results corresponding to the user behavior data in the computing power routing network, the computing power routing node can determine the next computing power routing node to which the first coding result or user traffic is transmitted. The specific determination process can be that the current computing power routing node can screen the multiple candidate nodes corresponding to the current computing power routing node to obtain the target candidate node, and determine the target candidate node as the next computing power routing node of the current computing power routing node.

[0062] Step 208, based on each target candidate node, generate an intent-aware path, and determine the computing power resources corresponding to the intent-aware path in the computing power resource pool.

[0063] The computing resource pool may include various types of computing resources, for example, computing resources of different models and computing resources from different manufacturers.

[0064] Specifically, after determining the target candidate node of the current computing power routing node, the system can determine the target candidate node as the next computing power routing node of the current computing power routing node, and transmit the first encoding result corresponding to the user behavior data and the user traffic to the next computing power routing node. In this way, the system can determine the next computing power routing node that receives the first encoding result corresponding to the user behavior data and the user traffic as the current computing power routing node, re-execute the steps after the above embodiment, obtain the next computing power routing node of the current computing power routing node, until the current computing power routing node reaches the computing power resource pool, and generate an intention-aware path. In this way, the system can determine the computing power resource corresponding to the intention-aware path based on the computing power routing node at the end position of the generated intention-aware path, among the multiple types of computing power resources contained in the computing power resource pool, and allocate the computing power resource to the user corresponding to the user behavior data and / or user traffic, and the user can communicate based on the computing power resource, wherein the computing power resource is a computing power resource that can meet the user's access needs and / or resource operation behavior.

[0065] In the above-mentioned routing method based on intent-aware paths, user behavior data is processed through a multi-dimensional intent recognition model to quickly and accurately obtain the first coding results corresponding to the user's target user intentions, and screen them in the computing power routing network to generate intent-aware paths. Based on the intent-aware paths, low-latency guidance of business routing is achieved, direct perception of computing power requirements and direct analysis of users' high-level requirements are achieved, thereby improving the resource utilization, security and performance of the computing power scheduling system, so that user needs can be responded to more quickly and accurately.

[0066] In an exemplary embodiment, Figure 3 As shown, the specific implementation process of the step of "determining the initial candidate node of the next node of the node; and screening each initial candidate node based on the node routing table of the node and the first encoding result to obtain the target candidate node" may include:

[0067] Step 302, based on the topological structure of the computing power routing network, determine the nodes that are directly connected to the node as initial candidate nodes for the next node of the node.

[0068] Among them, the topological structure of the computing power routing network represents the connection relationship between each computing power routing node contained in the computing power routing network.

[0069] Specifically, the system can screen out nodes that are directly connected to the current node (current computing power routing node) based on the topological structure of the computing power routing network, and determine that the nodes that are directly connected to the current node can be used as the initial candidate nodes for the next node of the current node. In other words, the system can screen among the multiple initial candidate nodes to determine the next node of the current node.

[0070] Step 304 : Determine, among the initial candidate nodes, based on the node routing table of the node, a second encoding result of the user intention satisfied by each initial candidate node.

[0071] Each node may store a node routing table, which may include multiple IP address information and encoding results corresponding to the IP address information, wherein the IP address information may be a complete IP address or prefix information of the IP address. The node routing table is used to determine the next node that meets the target user's intention. The encoding result corresponding to the IP address information may indicate that the IP address can meet the user's intention corresponding to the encoding result.

[0072] Specifically, the system can determine the IP address information corresponding to each initial candidate node, and can query the node routing table stored in the current node based on the IP address information of the initial candidate node to obtain the encoding result corresponding to each initial candidate node (hereinafter referred to as the second encoding result), as well as determine the user intentions that each initial candidate node can satisfy, and the second encoding results corresponding to the user intentions that each initial candidate node can satisfy.

[0073] Step 306 , screening is performed based on the second encoding results of the user intentions satisfied by each initial candidate node and a plurality of first encoding results corresponding to the user behavior data to obtain a target candidate node.

[0074] Specifically, the system can match the second encoding results of the user intentions that each initial candidate node can satisfy with the first encoding results corresponding to the user behavior data received by the current node, and screen out one or more target candidate nodes that can satisfy the target user's intentions.

[0075] Optionally, the system can determine multiple first encoding results of user behavior data received by the current node; the system can also determine the second encoding results of user intentions that each initial candidate node can satisfy, and match the multiple first encoding results corresponding to the target user intention with the second encoding results corresponding to the user intention that each initial candidate node can satisfy. If the second encoding result of the user intention that the initial candidate node can satisfy includes multiple first encoding nodes corresponding to the target user intention, the initial candidate node can be determined as the target candidate node.

[0076] Optionally, the multiple encoding results corresponding to the target user intent may include encoding result A and encoding result B. The system may first eliminate each initial candidate node that cannot satisfy the user intent of encoding result A based on the first encoding result of the user intent that can be satisfied by the initial candidate nodes respectively, and then eliminate each initial candidate node that cannot satisfy the user intent of encoding result B, and determine the target candidate node based on each retained initial candidate node, for example, each retained initial candidate node may be determined as the target candidate node.

[0077] In this embodiment, different user intentions are distinguished by encoding results, and the routing nodes are initially screened based on the encoding results to optimize the computing power routing path selection and improve the selection efficiency of the computing power routing nodes.

[0078] In one embodiment, Figure 4 As shown, the specific processing process of the step of "screening the second encoding results of the user intentions satisfied by each initial candidate node and the multiple first encoding results corresponding to the user behavior data to obtain the target candidate node" may include:

[0079] Step 402 , based on the multiple first coding results corresponding to the user behavior data, multiple initial candidate nodes are screened based on the second coding results satisfied by each initial candidate node to obtain each first candidate node matching the first coding result.

[0080] Specifically, the system can determine multiple first coding results corresponding to the user behavior data received by the current node; the system can also determine the second coding results of the user intent that can be satisfied by each initial candidate node, and match each initial candidate node with multiple first coding results corresponding to the target user intent. If the second coding results of the user intent that can be satisfied by the initial candidate node contain multiple first coding nodes corresponding to the target user intent, the initial candidate node is determined to be the first candidate node. Based on this, the system can perform the steps in this embodiment on each initial candidate node to obtain multiple first candidate nodes.

[0081] Step 404, based on the mapping relationship between the preset encoding results and the user intentions, determine the target user intentions corresponding to each first encoding result, and calculate the weight of the target user intentions with the index value corresponding to each target user intention in each first candidate node to obtain the cost value of each first candidate node.

[0082] Among them, the color coding result can be a multi-bit color coding, for example, it can be a 32-bit color coding. The mapping relationship between the preset color coding result and the user intention can be a mapping relationship between the dynamic color coding result and the user intention. Among them, each coding result can correspond to a different user intention. For example, a mapping relationship can be pre-configured between the coding result and the user intention. The user intention corresponding to the coding result A can be a low-latency network transmission, the user intention corresponding to the coding result B can be a low-cost network transmission, and the user intention corresponding to the third coding result C can be a request for computing resources, etc. Based on this, the system can determine the user intentions corresponding to each coding result in the output result, that is, the target user intention represented by the user behavior data, by querying the mapping relationship between the preset coding result and the user intention.

[0083] Optionally, the pre-configured weights of each user intention respectively represent the importance of different performance indicators. For example, the user intention may be to require low-latency network transmission, and the performance indicator corresponding to the user intention may be delay, and the corresponding indicator value may be the performance indicator value, that is, the delay value. The indicator value corresponding to the user intention satisfied by each first candidate node is the indicator value of the performance indicator related to the target user intention on the first candidate node. For example, the performance indicator related to the target user intention may be delay, and correspondingly, the indicator value corresponding to the user intention satisfied by the first candidate node may be the delay value on the first candidate node.

[0084] Specifically, the system can determine the performance indicators related to the target user's intention. After determining each first candidate node, for each candidate node, the system can collect the indicator values ​​corresponding to the performance indicators related to the target user's intention on the first candidate node, and calculate the indicator value and the weight of each performance indicator through a preset cost algorithm to obtain the cost value corresponding to the first candidate node.

[0085] Step 406 , screening is performed based on the cost value of each first candidate node to obtain a target candidate node.

[0086] Specifically, the system can calculate the index value corresponding to each first candidate node and the weight of the performance index based on a preset cost algorithm to obtain the cost value corresponding to each first candidate node. Based on this, the system can screen based on the cost value of each first candidate node to obtain the first candidate node that meets the preset screening condition, and determine the first candidate node that meets the preset screening condition as the target candidate node.

[0087] Optionally, the preset screening condition may be to screen out the node with the lowest cost value, or may be to screen out the target number of nodes with the lowest cost value. Correspondingly, the system may determine the first candidate node with the lowest cost value as the target candidate node. The system may also screen out the target number of first candidate nodes with the lowest cost value, and determine the first candidate node as the target candidate node.

[0088] Optionally, the system may calculate the cost value corresponding to the first candidate node by the following formula:

[0089]

[0090] Among them, the performance indicators related to the target user's intention may include delay, bandwidth and packet loss rate; P represents the path, and may also represent the node in the path, that is, the first candidate node; Cost represents the cost value, Delay represents the delay value on the first candidate node P; Bandwidth represents the bandwidth on the first candidate node P, and Loss represents the packet loss rate on the first candidate node P. w1 represents the weight of the delay, w2 represents the weight of the bandwidth, and w3 represents the weight of the packet loss rate.

[0091] In this embodiment, the costs corresponding to each first candidate node are calculated respectively by using pre-configured weights, so that multiple low-cost paths can be provided to users, better adapting to user needs and improving the efficiency of computing power routing.

[0092] In an exemplary embodiment, the intent-aware path includes multiple paths. Specifically, the system can generate multiple paths based on each target candidate node. For example, the target candidate nodes corresponding to the current node may include node 1 and node 2. Correspondingly, the system will determine the target candidate node corresponding to node 1, and then generate a first path. The system will also determine the target candidate node corresponding to node 2, and then generate a second path. In this way, the system can obtain an intent-aware path based on each path.

[0093] like Figure 5 As shown, the specific processing process of the step "determining the computing resources corresponding to the intent perception path in the computing resource pool" includes:

[0094] Step 502: In a computing resource pool containing multiple types of computing resources, determine the computing resources corresponding to the destination network endpoints of each path.

[0095] The computing resource pool may include various types of computing resources, for example, computing resources of different models and computing resources from different manufacturers.

[0096] Specifically, the system can determine the destination network ports corresponding to each path included in the intent-aware path, that is, determine the computing power routing node at the end of each path. For each path, the system can determine the computing power routing node at the end of the path, and based on the pre-configured correspondence between each computing power routing node and computing power resources, obtain the computing power resources corresponding to the computing power routing node at the end. Based on this, the system can obtain the computing power resources corresponding to each path.

[0097] Step 504, based on the priority of each type of computing resources, the computing resources corresponding to each destination network endpoint are screened to obtain the computing resources corresponding to the intent-aware path.

[0098] Among them, the priority of each type of computing resources can be pre-configured based on the actual application scenario.

[0099] Specifically, the system can screen the computing resources corresponding to the destination network endpoints of each path based on the pre-configured priorities of various types of computing resources, and extract the computing resources corresponding to the intent-aware path.

[0100] Optionally, the system can determine the computing power resource with the highest priority as the computing power resource corresponding to the intent-aware path. Alternatively, the system can determine the target computing power resource with the highest priority as the computing power resource corresponding to the intent-aware path, and output it to the user, who can select from the target computing power resource based on actual needs to obtain the computing power resource actually used to meet the user's needs.

[0101] In this embodiment, different path priorities are preconfigured to achieve automatic selection in a multi-intention and multi-path situation, thereby achieving more flexible path steering and load balancing.

[0102] In an exemplary embodiment, the routing method based on intent-aware path further includes:

[0103] User feedback data is obtained, and the mapping relationship is updated based on the user feedback data to obtain an updated mapping relationship.

[0104] Each mapping relationship is configured with version information; after obtaining the updated mapping relationship, the version information of the updated mapping relationship can also be updated accordingly, and backtracking can be performed based on the version information.

[0105] Specifically, the system can obtain data from real-time user feedback, update the mapping relationship between the preset color coding result and the user's intention, and obtain an updated mapping relationship. For example, the version information of the original mapping relationship can be the first version information, and the original mapping relationship is updated to obtain an updated mapping relationship. Correspondingly, the version information of the updated mapping relationship can be the second version information. The mapping relationship can be dynamic, for example, the mapping relationship can be stored through a variety of data storage mechanisms such as a database, and the mapping relationship can be adjusted based on the real-time data fed back by the user to obtain an updated new version of the mapping relationship, and each version of the mapping relationship is retained in the database.

[0106] In this embodiment, the mapping relationship can be adjusted in a timely manner to better meet the needs of actual application scenarios. The mapping relationship of each version can also be retained to facilitate backtracking, which is conducive to efficiently and accurately analyzing the source and reason of the change in the mapping relationship.

[0107] In an exemplary embodiment, the encoding result is a color encoding result; and the mapping relationship between the preset encoding result and the user's intention is the mapping relationship between the preset color encoding result and the user's intention. Figure 6 As shown, the routing method based on the intention-aware path also includes:

[0108] Step 602 : Collect the user's evaluation information on the target user's intention, and determine a color coding result that meets a preset obfuscation condition based on the evaluation information.

[0109] Among them, the user's evaluation information on the target user's intention can be the evaluation feedback information of the target user's intention corresponding to the user behavior data output by the system. For example, the user's evaluation information on the target user's intention can be the evaluation feedback information for each target user's intention, which can include the correct recognition intention or the wrong recognition intention; in the case where the evaluation information is the wrong recognition intention, the evaluation feedback information can also include the correct intention fed back by the user; the preset confusion condition can be the color coding result corresponding to the user's intention that appears the target number of times. The system can obtain the evaluation information fed back by the user in the form of a user log, a feedback form, etc.

[0110] Specifically, the system can output an evaluation request to the user when allocating computing resources to the user. In this way, the system can obtain evaluation information on the target user's intention from user feedback. If the evaluation information of the target user's intention shows that the number of times the intention is incorrectly recognized exceeds a preset threshold, the system can determine that the color coding result corresponding to the target user's intention is a color coding result that meets the preset confusion condition.

[0111] Step 604: based on the color coding results that meet the preset obfuscation conditions, improve the mapping relationship to obtain an improved mapping relationship.

[0112] Specifically, the system can adjust the mapping relationship between the preset color coding results and the user intentions that meets the preset confusion conditions based on the correct intention of the user feedback contained in the evaluation information of the incorrectly recognized intention. For example, the system can add a corresponding relationship between a new color coding result and a new user intention in the mapping relationship, or adjust the color coding result and the user intention, etc. Based on this, the system can obtain an improved mapping relationship.

[0113] In this embodiment, color coding results that have been misunderstood and confused multiple times can be quickly determined and adjusted in a timely manner to improve the accuracy of the mapping relationship and ensure the accuracy of the intent analysis.

[0114] The specific implementation process of the above-mentioned routing method based on intent-aware path is described in detail below in conjunction with a specific embodiment:

[0115] This embodiment provides a more flexible and intelligent computing power routing solution that can simultaneously perceive network status and computing resources to achieve more efficient resource management and optimization. This patent proposes a computing power routing method and device based on intent-aware paths, which uses BGP intent-aware paths to guide traffic for business routing that requires specific intents, thereby improving the resource utilization and security of the computing power scheduling system and optimizing the performance of the computing power scheduling system.

[0116] like Figure 7 As shown in the figure, it is the framework of the intent-aware computing power routing solution used in this embodiment, including users, network devices, computing power routing nodes and computing power resource pools. The network devices and computing power routing nodes communicate through the network. The network devices include user identity authentication modules, routing modules, computing power service routing encapsulation and intent analysis modules; the computing power routing nodes include computing resource perception modules, routing decision modules, BGP extension modules and computing power service routing identification modules, which can better meet the computing power network's needs in terms of resource scheduling and supply and demand balance. Specifically, as Figure 8 As shown, it may be a specific structural diagram of a computing power routing node, which includes a computing resource perception module, a routing decision module, a BGP extension module and a computing power service routing identification module. Resource scheduling is implemented through the computing resource perception module, routing decisions are implemented through the routing decision module, BGP extension is performed through the BGP extension module, and routing identification is performed through the computing power service routing identification module.

[0117] The intent parsing module is a key functional module of the intent-aware computing power routing solution framework. It can automatically analyze the user's high-level intentions. Specifically, it converts the user's intentions into specific network and computing power service requests by building a data model, and finally completes the routing encapsulation and passes it to the next node. The intent parsing module can automatically analyze the user's high-level intentions and convert them into specific computing power network requirements, reducing manual intervention and improving the intelligence level of computing power scheduling.

[0118] In one example, the specific process of intent-aware routing can be as follows: Fig. 9 As shown, the user is authenticated through the user identity authentication module, the user intention is analyzed through the intention parsing module, the routing request is processed through the computing power service routing encapsulation module, and the request is forwarded to the computing power routing node through the routing module.

[0119] The process of effectively converting user intent into network traffic routing requirements can include the following steps:

[0120] S1. User authentication and intention collection module; specifically, the identity is verified through the identity authentication module. After the verification is passed, the system identifies and records the computing network resource requirements and operation intentions of the user's access, for example, the user requests computing resources, hopes for low-latency network transmission, or requires lower costs, etc.; for example, the user identity can be confirmed through a user authentication service (such as OAuth, LDAP or SAML). After the verification is confirmed, the user's access requirements and operation behaviors can be recorded during the session. These operation behavior data are stored through the log or event tracking system, that is, the user behavior data is obtained.

[0121] S2. Intent perception: Specifically, the intent parsing module (intent perception module) analyzes the user's operation intention and converts it into specific network and computing power requirements. For example, it can parse user behavior data to build a data model and use natural language processing (NLP) and machine learning (ML) technologies to automatically identify the user's high-level intention. After analyzing the high-level intention, this module generates a computing resource request that matches the user's needs.

[0122] Optionally, a multidimensional data model based on user demand characteristics is constructed to map and parse different intents. Each dimension is constructed using a multidimensional data cube, and each dimension is regarded as an independent axis. Each axis represents a type of data feature. Such a data structure facilitates fast query and filtering of data in different dimensions. For example, a data cube can be constructed, in which each dimension represents user demand, service quality, cost, resource type, etc., so that slicing or aggregation can be performed on any combination of dimensions. Multidimensional Indexing: In order to support fast access to data of different dimensions, multidimensional indexes (such as R-tree, KD-tree, Bitmap Index, etc.) are introduced to achieve efficient data retrieval in high dimensions, which is suitable for querying specific intents such as low-latency paths and low-cost paths.

[0123] Optionally, the intent can be extracted and classified from the user behavior data through machine learning or deep learning models (such as LSTM, BERT, etc.), and the demand can be mapped to computing power and network resource requests through intent classification; in the model, the parsed intent is converted into a 32-bit color-coded result, and each color-coded result corresponds to a different user intent (such as low latency corresponds to color 1, low cost corresponds to color 2), and is assigned a corresponding network identifier. The mapping relationship between the color-coded result and the user intent can be updated, for example, a database or other data storage mechanism can be used to ensure that the mapping relationship can be automatically updated according to real-time data and user feedback; a version control mechanism is introduced in the mapping relationship, a new version is generated after each update, and the historical record is retained so that it can be traced back to the previous version when necessary, which helps to track the source and cause of the change; collect the user's behavior data when using the system and the evaluation information of the user feedback, obtain the color-coded result that meets the preset confusion condition, and then determine the mapping relationship that needs to be improved, and obtain the user's evaluation information based on user logs, feedback forms, etc. In addition, when the user's needs or intentions change, the system can quickly adjust the mapping relationship.

[0124] S3. BGP CACR (Color-Aware Computing Routing) mechanism; Specifically, the system can match computing service intent with a specific path by extending the SAFI (Subsequent Address Family Identifier) ​​in the BGP protocol. Each color-coded routing request is transmitted to the computing node according to the intent.

[0125] Optionally, two new SAFI types are defined in the BGP protocol, BGP CACR SAFI is used for network infrastructure path transmission, and VPN CACR SAFI is used to distinguish different intention routes of customer networks. A node routing table can be stored in each computing power routing node, and each routing table item in the node routing table contains an IP prefix and a color code. The router device selects and forwards the path according to the color identification, and forwards the user traffic to the next computing power routing node corresponding to the current computing power routing node, that is, the target candidate node. The intent-aware path generated by the above embodiment or the allocated computing power resources can meet the user's demand intention. Correspondingly, the BGP extension module can support multi-path strategies, and the color coding can be combined to define the traffic scheduling priority of different intentions. The intention is routed to different paths by color to achieve more accurate path selection, and it can also be tested and simulated in a preset small range in advance to ensure that the new mechanism can operate effectively in the existing infrastructure.

[0126] S4. Path selection and forwarding strategy; Specifically, each router node selects a path based on the color identifier configured by BGP CACR. The ingress PE or ASBR node guides the colored intent path to the computing power routing node and ensures that the user request reaches the optimal path. For example, using MPLS labels or SRv6 SIDs, combined with color coding, to set up a stack for the routing path, so that different intent paths are automatically resolved and the best path is selected in cross-domain delivery;

[0127] Optionally, each node verifies whether the path meets the intent (such as delay requirements) during forwarding. Through SR (Segment Routing) or Flex-Algo strategies, the path is automatically adjusted during the forwarding process. During the color verification process, Flex-Algo is used to select the optimal path based on the color identification of the user's intent, and adjust it in real time when the path conditions change. For example, when the delay of a path exceeds the preset threshold, Flex-Algo can dynamically select other paths that meet the conditions. Flex-Algo uses a weighted algorithm to calculate the cost value of the path:

[0128]

[0129] Among them, the performance indicators related to the target user's intention may include delay, bandwidth and packet loss rate; P represents the path, and may also represent the node in the path, that is, the first candidate node; Cost represents the cost value, Delay represents the delay value on the first candidate node P; Bandwidth represents the bandwidth on the first candidate node P, and Loss represents the packet loss rate on the first candidate node P. w1 represents the weight of the delay, w2 represents the weight of the bandwidth, and w3 represents the weight of the packet loss rate.

[0130] Alternatively, if Fig.10 As shown, the computing power routing network may include multiple areas, for example, domain 1, domain 2, and domain 3; the correspondence between the color coding results and user intent in each area may be different. The user intent of the service flow N1 received by the E1 interface may be low latency, and the user intent of the service flow N2 may be low latency and avoid certain resources. When routing and forwarding between multiple domains, the intent label of the cross-domain path is redefined through the color-to-intent mapping of the domain boundary to ensure consistent intent transmission between different network domains. In this way, the service flow N1 can be allocated to the computing power resource C1 through the interface E2, and the service flow N2 can be allocated to the computing power resource C2.

[0131] S5. Automatic service redirection mechanism; Specifically, according to the color label generated by the intent resolution module, each entry node will automatically redirect the service traffic to the color-aware path that meets the intent. At each forwarding hop, the system will perform path label exchange to achieve multi-level resolution and forwarding of labels and intents. Combined with IGP Flex-Algo and BGP CACR path priority, automatic path selection for multiple intents is achieved. Specify the path priority of different intents in the SR policy, combined with the BGP CACR path selection module, to achieve more flexible path steering and load balancing.

[0132] The routing based on the intent-aware path provided in this embodiment can decompose the routing strategy into different domains through flat and hierarchical design in a large-scale network. Each domain can realize the intention transmission of cross-domain routing through color boundary identification. The ingress BR reduces the consumption of core resources through the self-hierarchical design of the next hop. For example, a mapping table of different colors to intents can be set through a hierarchical routing structure to ensure the intention of cross-domain path selection. Figure 1 In other words, the service automatic steering mechanism can automatically redirect the service traffic to the path that meets the intent according to the color label generated by the intent parsing module, and use IGPFlex-Algo and BGP CACR path priority to achieve multi-intention automatic path selection. In a multi-domain environment, the routing strategy is decomposed through hierarchical design, and a mapping table of different colors to intents is set to ensure the intent of cross-domain path selection. Figure 1 The support of signaling protocol enables the path intentions between different color domains to be synchronized, thereby improving the utilization efficiency of network resources and service quality.

[0133] Based on this, the concept of color is introduced to distinguish different computing intentions or needs, and the BGP protocol is combined to optimize the computing power routing selection. By integrating functions such as user identity authentication module, intention analysis module and computing power service routing encapsulation, the efficiency of the computing power network in resource scheduling and supply and demand balance is effectively improved. The intention analysis module can automatically analyze the user's high-level needs and convert them into specific computing power and network service requests, thereby reducing manual intervention and reducing the complexity of resource allocation. Through this automated process, user needs can be responded to more quickly and accurately, significantly improving the utilization of computing power resources.

[0134] The color coding result in this embodiment is a 32-bit value associated with a specific user intent. The network device can package and parse according to the different intents of the user operation. By implementing intent-aware routing, users can directly express their specific needs for computing network resources after identity authentication. The system will automatically collect and parse these needs and convert them into executable requests, which can reduce the user's participation cost in the resource scheduling process and improve their overall experience. The system can intelligently process requests to ensure that network services can meet their needs in a timely manner.

[0135] In this embodiment, the BGP CACR (Color-Aware Computing Routing) mechanism is used to match the user's intention with a specific path, achieving more accurate routing selection. Each routing request is color-coded to correspond to the user's intention, and the router can select the optimal path based on the color identification. This intent-based routing configuration method enables the network to flexibly respond to the needs of different users while ensuring the optimal transmission effect of network traffic and reducing delays and costs.

[0136] It should be understood that, although the steps in the flowcharts involved in the above embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps is not strictly limited in order, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above embodiments may include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0137] Based on the same inventive concept, the embodiment of the present application also provides an intent-aware path-based routing device for implementing the above-mentioned intent-aware path-based routing method. The implementation solution provided by the device to solve the problem is similar to the implementation solution recorded in the above-mentioned method, so the specific limitations in one or more intent-aware path-based routing device embodiments provided below can refer to the limitations of the intent-aware path-based routing method above, and will not be repeated here.

[0138] In an exemplary embodiment, Fig.11 As shown, a routing device 1100 based on an intent-aware path is provided, comprising:

[0139] A first acquisition module 1102 is used to acquire user behavior data, where the user behavior data includes access requirements and / or resource operation behaviors;

[0140] A first determination module 1104 is used to process the user behavior data through a multi-dimensional intention recognition model to obtain a plurality of first encoding results corresponding to the user behavior data;

[0141] The second determination module 1106 is used to determine, for each node in the computing power routing network, an initial candidate node of the next node of the node; and to screen the initial candidate nodes based on the node routing table of the node and the first encoding result to obtain a target candidate node;

[0142] The third determination module 1108 is used to generate an intent-aware path based on each target candidate node, and determine the computing power resources corresponding to the intent-aware path in the computing power resource pool.

[0143] In one embodiment, the second determination module is specifically configured to:

[0144] Based on the topological structure of the computing power routing network, nodes that are directly connected to the node are determined as initial candidate nodes for the next node of the node;

[0145] In each initial candidate node, based on the node routing table of the node, determine a second encoding result of the user intention satisfied by each initial candidate node;

[0146] Based on the second encoding results of the user intentions satisfied by each initial candidate node and multiple first encoding results corresponding to the user behavior data, a target candidate node is obtained by screening.

[0147] In one embodiment, the second determination module is specifically configured to:

[0148] Based on the multiple first coding results corresponding to the user behavior data, screening is performed among the multiple initial candidate nodes based on the second coding results respectively satisfied by the initial candidate nodes to obtain the first candidate nodes matching the first coding results;

[0149] Based on the preset mapping relationship between the encoding result and the user intention, the target user intention corresponding to each first encoding result is determined, and the cost value of each first candidate node is obtained by calculating the weight of the target user intention and the index value corresponding to each target user intention in each first candidate node;

[0150] Screening is performed based on the cost value of each first candidate node to obtain a target candidate node.

[0151] In one embodiment, the intention perception path includes multiple paths; and the third determination module is specifically configured to:

[0152] In the computing resource pool containing multiple types of computing resources, the computing resources corresponding to the destination network endpoints of each path are determined respectively;

[0153] Based on the priority of each type of computing resources, the computing resources corresponding to each destination network endpoint are screened to obtain the computing resources corresponding to the intent-aware path.

[0154] In one embodiment, the device further comprises:

[0155] The second acquisition module is used to acquire user feedback data, and update the mapping relationship based on the user feedback data to obtain an updated mapping relationship, each mapping relationship being configured with version information.

[0156] In one embodiment, the encoding result is a color encoding result; the mapping relationship between the preset encoding result and the user intention is a mapping relationship between the preset color encoding result and the user intention; the device further includes:

[0157] A collection module, used to collect user evaluation information on the target user's intention, and determine a color coding result that meets a preset obfuscation condition based on the evaluation information;

[0158] The improvement module is used to improve the mapping relationship based on the color coding results that meet the preset confusion conditions to obtain an improved mapping relationship.

[0159] Each module in the above-mentioned routing device based on intention-aware path can be implemented in whole or in part by software, hardware and a combination thereof. Each of the above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to each of the above modules.

[0160] In an exemplary embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as shown in FIG. Fig.12 As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store intent data. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a routing method based on an intent-aware path is implemented.

[0161] Those skilled in the art will understand that Fig.12 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0162] In an exemplary embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the embodiments of the present application when executing the computer program.

[0163] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in the embodiments of the present application are implemented.

[0164] In one embodiment, a computer program product is provided, including a computer program, which implements the steps in the embodiments of the present application when executed by a processor.

[0165] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0166] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, an artificial intelligence (AI) processor, etc., but are not limited to this.

[0167] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0168] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A routing method based on intent-aware path, characterized in that: The method comprises: Acquiring user behavior data, wherein the user behavior data includes access requirements and / or resource operation behaviors; Processing the user behavior data through a multi-dimensional intention recognition model to obtain a plurality of first encoding results corresponding to the user behavior data; For each node in the computing power routing network, determine an initial candidate node for the next node of the node; and based on the node routing table of the node and the first encoding result, screen each of the initial candidate nodes to obtain a target candidate node; Based on each of the target candidate nodes, an intent-aware path is generated, and computing resources corresponding to the intent-aware path are determined in a computing resource pool.

2. The method according to claim 1, characterized in that The determining of the initial candidate node of the next node of the node; and screening each of the initial candidate nodes based on the node routing table of the node and the first encoding result to obtain a target candidate node, including: Based on the topological structure of the computing power routing network, nodes that are directly connected to the node are determined as initial candidate nodes for the next node of the node; In each of the initial candidate nodes, based on the node routing table of the node, determine a second encoding result of the user intention satisfied by each of the initial candidate nodes; Based on the second encoding results of the user intentions satisfied by each of the initial candidate nodes and the multiple first encoding results corresponding to the user behavior data, a target candidate node is obtained by screening.

3. The method according to claim 2, characterized in that The second encoding results based on the user intentions satisfied by each of the initial candidate nodes and the plurality of first encoding results corresponding to the user behavior data are screened to obtain the target candidate node, including: Based on the multiple first encoding results corresponding to the user behavior data, screening among the multiple initial candidate nodes based on the second encoding results respectively satisfied by the initial candidate nodes, to obtain first candidate nodes matching the first encoding results; Based on a preset mapping relationship between encoding results and user intents, determine the target user intents corresponding to the first encoding results, and calculate the cost value of each of the first candidate nodes by weighting the target user intents and the index value corresponding to each of the target user intents in the first candidate nodes; Screening is performed based on the cost value of each of the first candidate nodes to obtain a target candidate node.

4. The method according to claim 1, characterized in that: The intention perception path includes multiple paths; The determining, in the computing resource pool, computing resources corresponding to the intent-aware path includes: In the computing resource pool including multiple types of computing resources, respectively determine the computing resources corresponding to the destination network endpoints of each of the paths; Based on the priority of each type of computing power resources, the computing power resources corresponding to each of the destination network endpoints are screened to obtain the computing power resources corresponding to the intent-aware path.

5. The method according to claim 3, characterized in that: The method further comprises: User feedback data is obtained, and the mapping relationship is updated based on the user feedback data to obtain an updated mapping relationship, each of the mapping relationships being configured with version information.

6. The method according to claim 3, characterized in that: The encoding result is a color encoding result; the mapping relationship between the preset encoding result and the user intention is a mapping relationship between the preset color encoding result and the user intention; The method further comprises: Collecting user evaluation information on the target user's intention, and determining a color coding result that meets a preset obfuscation condition based on the evaluation information; Based on the color coding results that meet the preset confusion conditions, the mapping relationship is improved to obtain an improved mapping relationship.

7. A routing device based on intention-aware path, characterized in that: The device comprises: A first acquisition module, used to acquire user behavior data, wherein the user behavior data includes access requirements and / or resource operation behaviors; A first determination module, configured to process the user behavior data through a multi-dimensional intention recognition model to obtain a plurality of first encoding results corresponding to the user behavior data; A second determination module is used to determine, for each node in the computing power routing network, an initial candidate node of the next node of the node; and to screen each of the initial candidate nodes based on the node routing table of the node and the first encoding result to obtain a target candidate node; The third determination module is used to generate an intention-aware path based on each of the target candidate nodes, and determine the computing power resources corresponding to the intention-aware path in the computing power resource pool.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images