CDN acceleration strategy generation method, generation device and electronic equipment

Abstract

This application discloses a method, apparatus, and electronic device for generating a CDN acceleration strategy. The generation method includes: receiving a CDN acceleration strategy generation request sent by a target terminal, wherein the CDN acceleration strategy generation request carries a target tag combination, the target tag combination including a scenario tag, an application type tag, and a target user group tag; and generating a target CDN acceleration strategy corresponding to the target tag combination in response to the CDN acceleration strategy generation request.

Description

Technical Field

[0001] This application relates to the field of Content Delivery Network (CDN), and more particularly to a method, apparatus and electronic device for generating CDN acceleration strategies. Background Technology

[0002] With the continuous development of mobile internet technology and the arrival of the 5G+ era, the number of users and inter-network traffic in the mobile internet environment are gradually increasing. Operators can use CDN acceleration for streaming media, images and other resources to bring users a better mobile internet experience.

[0003] Currently, CDN acceleration strategies are mainly generated manually, which is inefficient. Summary of the Invention

[0004] This application discloses a method, apparatus, and electronic device for generating CDN acceleration strategies, which can improve the efficiency of CDN acceleration strategy generation.

[0005] To solve the above problems, this application adopts the following technical solution:

[0006] In a first aspect, embodiments of this application disclose a method for generating a CDN acceleration strategy, comprising: receiving a CDN acceleration strategy generation request sent by a target terminal, wherein the CDN acceleration strategy generation request carries a target tag combination, the target tag combination including a scenario tag, an application type tag, and a target user group tag; and generating a target CDN acceleration strategy corresponding to the target tag combination in response to the CDN acceleration strategy generation request.

[0007] Secondly, embodiments of this application disclose a CDN acceleration strategy generation apparatus, comprising: a receiving module, configured to receive a CDN acceleration strategy generation request sent by a target terminal, wherein the CDN acceleration strategy generation request carries a target tag combination, the target tag combination including a scenario tag, an application type tag, and a target user group tag; and a generation module, configured to generate a target CDN acceleration strategy corresponding to the target tag combination in response to the CDN acceleration strategy generation request.

[0008] Thirdly, embodiments of this application provide an electronic device including a processor and a memory, wherein the memory stores programs or instructions executable on the processor, and the programs or instructions, when executed by the processor, implement the steps of the method described in the first aspect.

[0009] Fourthly, embodiments of this application provide a readable storage medium on which a program or instructions are stored, which, when executed by a processor, implement the steps of the method described in the first aspect.

[0010] Fifthly, embodiments of this application provide a computer program product, the computer program product including a computer program stored on a non-transitory computer-readable storage medium, the computer program including program instructions, which, when executed by a computer, cause the computer to perform the steps of the method described in the first aspect.

[0011] The technical solution adopted in this application can achieve the following beneficial effects:

[0012] This application provides a method for generating a CDN acceleration strategy. The method involves receiving a CDN acceleration strategy generation request from a target terminal through a CDN scheduling platform. The CDN acceleration strategy generation request carries a target tag combination, which includes a scenario tag, an application type tag, and a target user group tag. Then, in response to the received CDN acceleration strategy generation request, the CDN scheduling platform generates a target CDN acceleration strategy corresponding to the target tag combination. Because this application generates the CDN acceleration strategy corresponding to the target tag combination through a CDN scheduling platform, it improves the efficiency of CDN acceleration strategy generation compared to manual decision-making. Attached Figure Description

[0013] Figure 1 This is a flowchart illustrating a method for generating a CDN acceleration strategy disclosed in an embodiment of this application;

[0014] Figure 2 This is a schematic diagram of the structure of a target acceleration strategy generation model disclosed in an embodiment of this application;

[0015] Figure 3 This is a flowchart illustrating a method for generating a CDN acceleration strategy as disclosed in an embodiment of this application;

[0016] Figure 4 This is a schematic diagram of the structure of a CDN acceleration strategy generation device disclosed in an embodiment of this application;

[0017] Figure 5 This is a schematic diagram of the structure of an electronic device disclosed in an embodiment of this application. Detailed Implementation

[0018] The technical solutions of the embodiments of this application will be clearly described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application are within the scope of protection of this application.

[0019] The terms "first," "second," etc., used in the specification and claims of this application are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such terms can be used interchangeably where appropriate so that embodiments of this application can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first," "second," etc., are generally of the same class and the number of objects is not limited; for example, a first object can be one or more. Furthermore, in the specification and claims, "and / or" indicates at least one of the electrically connected objects, and the character " / " generally indicates that the preceding and following objects are in an "or" relationship.

[0020] The CDN acceleration strategy generation method, generation device, and electronic device disclosed in this application will be described in detail below with reference to the accompanying drawings and through specific embodiments and application scenarios.

[0021] This application discloses a method for generating CDN acceleration strategies, which is applied to a CDN scheduling platform. Figure 1 This is a flowchart illustrating a method for generating a CDN acceleration strategy disclosed in an embodiment of this application. Figure 1 As shown, the method includes the following steps:

[0022] S120. Receive a CDN acceleration strategy generation request sent by the target terminal, wherein the CDN acceleration strategy generation request carries a target tag combination, the target tag combination including a scenario tag, an application type tag, and a target user group tag.

[0023] The target terminal in this application can be a CDN application provider.

[0024] For example, scene tags can include main scene tags and sub-scene tags. For instance, main scene tags can include long-form video movies and TV shows, short-form live streaming, etc., while sub-scene tags can include live broadcasts of major sporting events, online public beta tests of major updates, important marketing activities, etc. Application (APP) type tags can include news and information, film and television videos, public services, interactive games, etc. Target user group tags can include IT (Internet Technology) professionals, students, etc.

[0025] It should be noted that CDN's multi-tag acceleration strategy aims to achieve more refined and flexible resource allocation and optimization by assigning specific tags to different content or services. This strategy can help CDN distribute content more efficiently and improve user experience.

[0026] S140. In response to the CDN acceleration strategy generation request, generate a target CDN acceleration strategy corresponding to the target tag combination.

[0027] When the CDN scheduling platform receives a CDN acceleration policy generation request carrying a target combination tag from the target terminal, it generates a target CDN acceleration policy corresponding to the target tag combination and returns the generated target CDN acceleration policy to the target terminal. The tag combination and the CDN acceleration policy are in one-to-one correspondence.

[0028] For example, CDN acceleration strategies may include transmission configuration data (VPN (Virtual Private Network), dynamic acceleration, cloud-edge channel), configuration space allocation, uplink and downlink bandwidth, multi-level caching capabilities (mechanical disk, solid-state disk, memory), access schemes (DNS (Domain Name System), 302, HTTP DNS), etc.

[0029] This application provides a method for generating a CDN acceleration strategy. The method involves receiving a CDN acceleration strategy generation request from a target terminal through a CDN scheduling platform. The CDN acceleration strategy generation request carries a target tag combination, which includes a scenario tag, an application type tag, and a target user group tag. Then, in response to the received CDN acceleration strategy generation request, the CDN scheduling platform generates a target CDN acceleration strategy corresponding to the target tag combination. Because this application generates the CDN acceleration strategy corresponding to the target tag combination through a CDN scheduling platform, it improves the efficiency of CDN acceleration strategy generation compared to manual decision-making.

[0030] In this embodiment, generating a target CDN acceleration strategy corresponding to the target tag combination may include: determining a target acceleration strategy generation model corresponding to the target tag combination; and obtaining a target CDN acceleration strategy output by the target acceleration strategy generation model corresponding to the target tag combination by inputting the target tag combination into the target acceleration strategy generation model. The target acceleration strategy generation model is used to generate a CDN acceleration strategy corresponding to the input tag combination. In other words, this application generates a target CDN acceleration strategy corresponding to the target tag combination based on the acceleration strategy generation model, which can improve the generation efficiency and accuracy of CDN acceleration strategies.

[0031] In one implementation, determining the target acceleration strategy generation model corresponding to the target tag combination may include: if the target tag combination belongs to a pre-stored tag combination, determining the target acceleration strategy generation model corresponding to the target tag combination from an acceleration strategy generation model library.

[0032] The CDN scheduling platform pre-stores the correspondence between tag combinations and acceleration strategy generation models. Multiple tag combinations correspond to one acceleration strategy generation model, and the acceleration strategy generation model library can store multiple different acceleration strategy generation models.

[0033] When a CDN scheduling platform receives a CDN acceleration strategy generation request carrying a target tag combination, it determines whether the target tag combination belongs to the pre-stored tag combinations based on the tag combinations pre-stored by the CDN scheduling platform. If the target tag combination belongs to the pre-stored tag combinations, it determines the target acceleration strategy generation model corresponding to the target tag combination from the acceleration strategy generation model library. Then, by inputting the target tag combination into the target acceleration strategy generation model, it obtains the target CDN acceleration strategy corresponding to the target tag combination output by the target acceleration strategy generation model.

[0034] In another implementation, determining the target acceleration strategy generation model corresponding to the target label combination may include: creating a target acceleration strategy generation model to be trained when the target label combination does not belong to a pre-stored label combination, wherein the target acceleration strategy generation model consists of an input layer, an embedding layer, a general sub-model, a fully connected layer, a dropout layer, and an output layer, and the general sub-model is the common part of each acceleration strategy generation model in the acceleration strategy generation model library; acquiring multiple training samples, wherein the training samples include sample label combinations and sample CDN acceleration strategies corresponding to the sample label combinations, and the sample label combinations include at least two of the target label combinations; iteratively training the target acceleration strategy generation model to be trained based on the multiple training samples until the loss function corresponding to the target acceleration strategy generation model converges.

[0035] In this application, when the CDN scheduling platform receives a CDN acceleration strategy generation request carrying a target tag combination, it determines whether the target tag combination belongs to the pre-stored tag combinations based on the tag combinations pre-stored by the CDN scheduling platform. If the target tag combination does not belong to the pre-stored tag combinations, a target acceleration strategy generation model to be trained is created, and the target acceleration strategy generation model to be trained is iteratively trained based on training samples until the loss function corresponding to the target acceleration strategy generation model converges.

[0036] In this application, the multiple training samples used to train the target acceleration strategy generation model may include historical data and / or labeled data. The sample label combination in the training samples includes at least two of the target label combinations, thereby enabling the trained target acceleration strategy generation model to more accurately generate target CDN acceleration strategies corresponding to the target label combinations.

[0037] For example, if the main scene tag in the scene tags included in the target tag combination is long-form video / film, the sub-scene tag in the scene tags is live broadcast of a major sporting event, the application type tag is news / information, and the target user group tag is IT professionals, then the main scene tag in the scene tags included in the sample tag combination is long-form video / film, the sub-scene tag in the scene tags is live broadcast of a major sporting event, the application type tag is public service, and the target user group tag is IT professionals. Alternatively, the main scene tag in the scene tags included in the sample tag combination is long-form video / film, the sub-scene tag in the scene tags is live broadcast of a major sporting event, the application type tag is news / information, and the target user group tag is students. In addition, the sample tag combination may also include at least two of the target tag combinations, which will not be listed here.

[0038] The structure of the target acceleration strategy generation model is as follows: Figure 2As shown, the input layer of the target acceleration strategy generation model is a multi-label input layer used for input scene labels, application type labels, and target user group labels. The embedding layer of the target acceleration strategy generation model is used to convert the input words into 128-dimensional spatial vectors. When the input data dimension of the embedding layer is Multi-tag_vocab_size and the input sequence length is Multi-tag_length, the shape of the output data of the embedding layer is (None, Multi-tag_length, 128). The general sub-model of the target acceleration strategy generation model can be LLM (Large Language Model). The model (large language model) is a general CDN acceleration strategy generation model. The general sub-model is the common part of various acceleration strategy generation models in the acceleration strategy generation model library. The fully connected layer (Dense) of the target acceleration strategy generation model contains 64 Dense fully connected neurons, and the activation function is set to "relu". The fully connected layer connects each node to all nodes in the previous layer, and integrates the previously extracted features. The dropout layer of the target acceleration strategy generation model discards input data according to the dropout ratio to prevent the model from overfitting. For example, the dropout ratio can be set to 0.3. The output layer of the target acceleration strategy generation model is a fully connected layer, containing Dense fully connected neurons of the number Strategy_vocab_size, and the activation function is set to "softmax". The softmax output result is fed into the multi-class cross-entropy loss function. The shape of the output data of the output layer is (None, Strategy_vocab_size).

[0039] In this application, the weights of the general sub-model are used as the initial weights of the target acceleration policy generation model to be trained, thereby quickly generating a new target acceleration policy generation model.

[0040] Large-scale language models (LLMs) are a class of natural language processing (NLP) techniques based on deep learning. Their primary purpose is to enable machines to better understand and generate human-language text, such as articles, dialogues, and searches. LLMs typically require deep learning models, and the training data often reaches or exceeds one billion words. Due to the need for powerful computing resources and machine learning algorithms, their core idea is to allow machines to simulate human language processing capabilities, gradually learning and mastering vast amounts of language patterns, grammatical rules, and semantic knowledge, and then self-adjusting and optimizing for specific tasks. Currently, LLMs have been applied in multiple fields, such as chatbots, machine translation, information extraction, and sentiment analysis, and have become one of the hottest research topics in the field of natural language processing. The emergence of LLMs represents a significant step forward for machines in understanding and processing natural language text, and also helps improve cross-cultural communication and understanding between humans and machines.

[0041] The sample scene label set obtained based on multiple training samples can be represented as: V={v1,v2,...,v...} n}, where v n The feature vector of the nth word, and the set of application type labels obtained based on multiple training samples, can be represented as: Q = {q1, q2, ..., q} t}, where q t Let S be the feature vector of the t-th word. The target user group label set obtained based on multiple training samples can be represented as: S = {s1, s2, ..., s...} m}, where s m The feature vector of the m-th word, and the CDN acceleration strategy set obtained based on multiple training samples, can be represented as: Y = {y1, y2, ..., y} L}, where y L It is the feature vector of the l-th word.

[0042] Before iteratively training the target acceleration strategy generation model based on multiple training samples, the training samples can be preprocessed according to preset rules, and then the preprocessed training samples can be used for iterative training to speed up the training process.

[0043] Preprocessing of training samples may include: (1) cleaning and serializing the training samples. Specifically, unify the case of letters, convert uppercase letters to lowercase letters, remove all punctuation marks, then segment the text into words, and tokenize each word so that each piece of text is converted into a piece of index number, and pad the sequence with zeros if it does not reach the maximum text length. (2) Take the longest length Multi-tag_length of the sample scene tag set, sample application type tag set and sample target user group tag set as its unified index sequence length, and set the dictionary size of the sample scene tag set, sample application type tag set and sample target user group tag set to Multi-tag_vocab_size; take the longest length Strategy_length of the sample CDN acceleration strategy set as its unified index sequence length, and set the dictionary size of the sample CDN acceleration strategy set to Strategy_vocab_size.

[0044] Alternatively, 90% of the multiple training samples can be used as the training set to train the model, and 10% can be used as the test set to test the performance of the trained model.

[0045] For example, during model training, the number of training epochs can be set to 1000 (epochs = 1000), the batch size to 100 (batch_size = 100), and categorical cross-entropy can be chosen as the loss function, i.e., the objective function (loss = 'categorical_crossentropy'). The Adam optimizer can be selected to improve the learning speed of traditional gradient descent (optimizer = 'adam'). The neural network can find the optimal weight values ​​that minimize the objective function through gradient descent, and the neural network will autonomously learn these weight values ​​through training. The weights of the model are derived after convergence.

[0046] In this embodiment, the iterative training of the target acceleration policy generation model based on multiple training samples until the loss function corresponding to the target acceleration policy generation model converges may include: after freezing the weight parameters of the general sub-model, iteratively training the target acceleration policy generation model based on multiple training samples until the loss function corresponding to the target acceleration policy generation model converges. In other words, this application only trains the input layer, embedding layer, fully connected layer, dropout layer, and output layer. Transfer learning can accelerate the model's learning efficiency, and the scheme of this application can reduce the amount of training data used.

[0047] In one implementation, before obtaining the target CDN acceleration strategy corresponding to the target tag combination by inputting the target tag combination into the target acceleration strategy generation model, the method may further include: preprocessing the target tag combination according to a preset rule; the step of obtaining the target CDN acceleration strategy corresponding to the target tag combination by inputting the target tag combination into the target acceleration strategy generation model may include: obtaining the target CDN acceleration strategy corresponding to the target tag combination by inputting the preprocessed target tag combination into the target acceleration strategy generation model.

[0048] It should be noted that the specific implementation of the preprocessing of the target label combination here is the same as the preprocessing of the training samples described above, and will not be repeated here.

[0049] By inputting the preprocessed target tag combination into the target acceleration strategy generation model, the speed at which the target acceleration strategy generation model outputs the target CDN acceleration strategy corresponding to the target tag combination can be accelerated.

[0050] This application discloses a method for generating CDN acceleration strategies, such as... Figure 3 As shown, the CDN scheduling platform receives a CDN acceleration strategy generation request sent by the CDN application. This request carries a target tag combination. If the target tag combination belongs to a pre-stored tag combination (an existing tag combination), the platform determines the target acceleration strategy generation model corresponding to the target tag combination from the acceleration strategy generation model library. By inputting the target tag combination into the target acceleration strategy generation model, the platform obtains the target CDN acceleration strategy output by the model corresponding to the target tag combination and returns it to the CDN application. If the target tag combination does not belong to a pre-stored tag combination (an existing tag combination), the platform creates a target acceleration strategy generation model to be trained. After preprocessing the training samples through a preprocessing module, the platform iteratively trains the target acceleration strategy generation model based on the preprocessed training samples. Then, the platform inputs the target tag combination into the trained target acceleration strategy generation model to obtain the target CDN acceleration strategy output by the model corresponding to the target tag combination and returns it to the CDN application.

[0051] The CDN acceleration strategy generation method provided in this application can be executed by a CDN acceleration strategy generation device. This application uses an example of a CDN acceleration strategy generation device executing the CDN acceleration strategy generation method to illustrate the apparatus of the CDN acceleration strategy generation method provided in this application.

[0052] Figure 4 This is a schematic diagram of the structure of a CDN acceleration strategy generation device disclosed in an embodiment of this application. Figure 4 As shown, the CDN acceleration strategy generation device 400 includes a receiving module 410 and a generation module 420.

[0053] In this application, the receiving module 410 is used to receive a CDN acceleration strategy generation request sent by the target terminal, wherein the CDN acceleration strategy generation request carries a target tag combination, the target tag combination including a scenario tag, an application type tag, and a target user group tag; the generating module 420 is used to generate a target CDN acceleration strategy corresponding to the target tag combination in response to the CDN acceleration strategy generation request.

[0054] In one implementation, the generation module 420 generates a target CDN acceleration strategy corresponding to the target tag combination, including: determining a target acceleration strategy generation model corresponding to the target tag combination; and obtaining a target CDN acceleration strategy output by the target acceleration strategy generation model corresponding to the target tag combination by inputting the target tag combination into the target acceleration strategy generation model, wherein the target acceleration strategy generation model is used to generate a CDN acceleration strategy corresponding to the input tag combination.

[0055] In one implementation, the generation module 420 determines the target acceleration strategy generation model corresponding to the target tag combination, including: if the target tag combination belongs to a pre-stored tag combination, determining the target acceleration strategy generation model corresponding to the target tag combination from the acceleration strategy generation model library.

[0056] In one implementation, the generation module 420 determines a target acceleration strategy generation model corresponding to the target label combination, including: when the target label combination does not belong to a pre-stored label combination, creating a target acceleration strategy generation model to be trained, wherein the target acceleration strategy generation model consists of an input layer, an embedding layer, a general sub-model, a fully connected layer, a dropout layer, and an output layer, and the general sub-model is the common part of various acceleration strategy generation models in the acceleration strategy generation model library; acquiring multiple training samples, wherein the training samples include sample label combinations and sample CDN acceleration strategies corresponding to the sample label combinations, and the sample label combinations include at least two of the target label combinations; iteratively training the target acceleration strategy generation model to be trained based on the multiple training samples until the loss function corresponding to the target acceleration strategy generation model converges.

[0057] In one implementation, the generation module 420 iteratively trains the target acceleration strategy generation model to be trained based on multiple training samples until the loss function corresponding to the target acceleration strategy generation model converges, including: after freezing the weight parameters of the general sub-model, iteratively training the target acceleration strategy generation model to be trained based on multiple training samples until the loss function corresponding to the target acceleration strategy generation model converges.

[0058] In one implementation, the generation module 420 is further configured to preprocess the target tag combination according to a preset rule before obtaining the target CDN acceleration strategy output by the target acceleration strategy generation model corresponding to the target tag combination by inputting the target tag combination into the target acceleration strategy generation model; the generation module 420 obtaining the target CDN acceleration strategy output by the target acceleration strategy generation model corresponding to the target tag combination by inputting the target tag combination into the target acceleration strategy generation model includes: obtaining the target CDN acceleration strategy output by the target acceleration strategy generation model corresponding to the target tag combination by inputting the preprocessed target tag combination into the target acceleration strategy generation model.

[0059] The CDN acceleration strategy generation apparatus provided in this application embodiment can implement all the processes implemented in the CDN acceleration strategy generation method embodiment, and will not be described again here to avoid repetition.

[0060] like Figure 5 As shown, this application embodiment also provides an electronic device 500, including a processor 501 and a memory 502. The memory 502 stores a program or instructions that can run on the processor 501. When the program or instructions are executed by the processor 501, they implement the various steps of the above-described CDN acceleration strategy generation method embodiment and can achieve the same technical effect. To avoid repetition, they will not be described again here.

[0061] It should be noted that the electronic devices in the embodiments of this application include mobile electronic devices and non-mobile electronic devices.

[0062] This application also provides a readable storage medium storing a program or instructions. When the program or instructions are executed by a processor, they implement the various processes of the above-described CDN acceleration strategy generation method embodiment and achieve the same technical effect. To avoid repetition, they will not be described again here.

[0063] The processor is the processor in the electronic device described in the above embodiments. The readable storage medium includes computer-readable storage media, such as computer read-only memory (ROM), random access memory (RAM), magnetic disk, or optical disk.

[0064] This application also provides a computer program product, which includes a computer program stored on a non-transitory computer-readable storage medium. The computer program includes program instructions that, when executed by a computer, cause the computer to perform the steps of the CDN acceleration strategy generation method described above.

[0065] The above embodiments of this application focus on describing the differences between the various embodiments. As long as the different optimization features between the various embodiments are not contradictory, they can be combined to form a better embodiment. For the sake of brevity, they will not be described in detail here.

[0066] The above description is merely an embodiment of this application and is not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A method for generating a CDN acceleration strategy, characterized in that, include: Receive a CDN acceleration policy generation request sent by the target terminal, wherein the CDN acceleration policy generation request carries a target tag combination, the target tag combination including scenario tag, application type tag and target user group tag; In response to the CDN acceleration strategy generation request, a target CDN acceleration strategy corresponding to the target tag combination is generated; The generation of the target CDN acceleration strategy corresponding to the target tag combination includes: Determine the target acceleration strategy generation model corresponding to the target label combination; By inputting the target tag combination into the target acceleration strategy generation model, a target CDN acceleration strategy corresponding to the target tag combination is obtained from the output of the target acceleration strategy generation model. The target acceleration strategy generation model is used to generate a CDN acceleration strategy corresponding to the input tag combination, and the target CDN acceleration strategy is used to assign specific tags to different content or services.

2. The generation method according to claim 1, characterized in that, The determination of the target acceleration strategy generation model corresponding to the target label combination includes: If the target tag combination belongs to a pre-stored tag combination, the target acceleration strategy generation model corresponding to the target tag combination is determined from the acceleration strategy generation model library.

3. The generation method according to claim 1, characterized in that, The determination of the target acceleration strategy generation model corresponding to the target label combination includes: If the target label combination does not belong to the pre-stored label combination, a target acceleration policy generation model to be trained is created. The target acceleration policy generation model consists of an input layer, an embedding layer, a general sub-model, a fully connected layer, a dropout layer, and an output layer. The general sub-model is the common part of each acceleration policy generation model in the acceleration policy generation model library. Multiple training samples are obtained, wherein the training samples include sample label combinations and sample CDN acceleration strategies corresponding to the sample label combinations, and the sample label combinations include at least two of the target label combinations; The target acceleration policy generation model is iteratively trained based on multiple training samples until the loss function corresponding to the target acceleration policy generation model converges.

4. The generation method according to claim 3, characterized in that, The step of iteratively training the target acceleration policy generation model based on multiple training samples until the loss function corresponding to the target acceleration policy generation model converges includes: After freezing the weight parameters of the general sub-model, the target acceleration policy generation model to be trained is iteratively trained based on multiple training samples until the loss function corresponding to the target acceleration policy generation model converges.

5. The generation method according to claim 1, characterized in that, Before obtaining the target CDN acceleration strategy corresponding to the target tag combination output by the target acceleration strategy generation model by inputting the target tag combination into the target acceleration strategy generation model, the method further includes: The target tag combination is preprocessed according to preset rules; The step of inputting the target tag combination into the target acceleration strategy generation model to obtain the target CDN acceleration strategy output by the target acceleration strategy generation model corresponding to the target tag combination includes: By inputting the preprocessed target tag combination into the target acceleration strategy generation model, the target CDN acceleration strategy output by the target acceleration strategy generation model corresponding to the target tag combination is obtained.

6. A CDN acceleration strategy generation apparatus, characterized in that, include: The receiving module is used to receive a CDN acceleration strategy generation request sent by the target terminal, wherein the CDN acceleration strategy generation request carries a target tag combination, the target tag combination including a scenario tag, an application type tag and a target user group tag; The generation module is used to generate a target CDN acceleration strategy corresponding to the target tag combination in response to the CDN acceleration strategy generation request. The generation module is also used to determine the target acceleration strategy generation model corresponding to the target label combination; By inputting the target tag combination into the target acceleration strategy generation model, a target CDN acceleration strategy corresponding to the target tag combination is obtained from the output of the target acceleration strategy generation model. The target acceleration strategy generation model is used to generate a CDN acceleration strategy corresponding to the input tag combination, and the target CDN acceleration strategy is used to assign specific tags to different content or services.

7. An electronic device, characterized in that, It includes a processor and a memory, the memory storing a program or instructions that can run on the processor, the program or instructions being executed by the processor to implement the steps of the CDN acceleration strategy generation method as described in any one of claims 1-5.

8. A readable storage medium, characterized in that, The program or instructions are stored on the readable storage medium, and when the program or instructions are executed by the processor, they implement the steps of the CDN acceleration strategy generation method as described in any one of claims 1-5.

9. A computer program product, characterized in that, The computer program product includes a computer program stored on a non-transitory computer-readable storage medium, the computer program including program instructions that, when executed by a computer, cause the computer to perform the steps of the CDN acceleration strategy generation method as described in any one of claims 1-5.

Images