Distributed cache preheating loading method based on machine learning

Abstract

The invention discloses a distributed cache preheating loading method based on machine learning, belongs to the technical field of cache preheating, and can analyze hot data of a system in advance, identify the hot data in the initialization and operation process of the system, and preload the hot data in advance. In the distributed cache preheating loading method, machine learning is introduced, hot spot data of the system are recognized based on user behaviors and system logs, hot spot cache data are actively loaded, and therefore the system performance is further improved, and the user experience is improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of cache preheating, and in particular relates to a distributed cache preheating loading method based on machine learning. Background Art

[0002] A distributed system is a system composed of a group of independent computers that are connected through a network and work together to complete tasks. These computers can be physically dispersed or virtual, and they exchange information and work together through a communication network. The purpose of a distributed system is to improve the reliability, scalability, fault tolerance, and resource utilization of the system. Reasonable use of cache can significantly improve the query response speed in a distributed system and improve the user experience. However, in the process of using cache, when the system is initialized, there will be a situation where the system has no cache at all, resulting in a slow startup of the system, or even a failure to start due to excessive query requests. Or sudden hot events may cause cache breakdown. Summary of the invention

[0003] The present invention provides a distributed cache preheating loading method based on machine learning, which is used to solve the technical problems existing in the prior art.

[0004] A distributed cache preheating loading method based on machine learning, comprising:

[0005] Using system log data; wherein the system log data is historical data stored in a distributed system;

[0006] Parsing the system log data to determine historical user behavior data; wherein the historical user behavior data includes user behavior and user behavior time;

[0007] A distributed cache preheating loading model is constructed using a machine learning algorithm, and historical user behavior data is learned through the distributed cache preheating loading model to obtain a distributed cache preheating loading model with predictive capabilities;

[0008] Collect real-time user behavior data, and analyze the real-time user behavior data through a distributed cache preheating loading model with prediction capabilities to determine the behavior analysis results;

[0009] Based on the behavior analysis results, the data in the distributed system is cached to complete the distributed cache preheating loading based on machine learning.

[0010] Furthermore, the system log data is used, including: connecting to the distributed system through an API interface, and obtaining the system log data stored in the distributed system.

[0011] Furthermore, a distributed cache preheating loading model is constructed using a machine learning algorithm, and historical user behavior data is learned through the distributed cache preheating loading model to obtain a distributed cache preheating loading model with predictive capabilities, including:

[0012] Use machine learning algorithms to build a distributed cache preheating loading model;

[0013] Based on the historical user behavior data, determine N historical user behavior data at a historical time point as sample data, and determine the N+1th historical user behavior data as expected data;

[0014] The distributed cache preheating loading model is used to learn the association relationship between the sample data and the expected data, thereby obtaining a distributed cache preheating loading model with prediction capability.

[0015] Furthermore, the distributed cache preheating loading model is used to learn the association between the sample data and the expected data, so as to obtain a distributed cache preheating loading model with prediction capability, including:

[0016] Initialize the model parameters corresponding to the distributed cache preheating loading model, where the model parameters are weight parameters;

[0017] Using the sample data as input of a distributed cache preheating loading model, obtaining actual output data of the distributed cache preheating loading model;

[0018] According to the expected data and the actual output data, an error function value corresponding to the distributed cache preheating loading model is obtained;

[0019] It is determined whether the error function value is less than a preset threshold. If so, a distributed cache preheating loading model with prediction capability is obtained. Otherwise, the weight of the distributed cache preheating loading model is updated and the next training is started.

[0020] Furthermore, the model parameters corresponding to the distributed cache preheating loading model are initialized, including:

[0021] Between the upper and lower limits of the parameters of the distributed cache preheating loading model, a random initialization method or a chaotic mapping initialization method is used to generate model parameters to achieve parameter initialization.

[0022] Furthermore, the error function value is obtained by an error function, and the error function is:

[0023]

[0024] Where L represents the error function, p = 1, 2, .., P, P represents the total number of input data, k = 1, 2, .., K, K represents the total number of neurons in the output layer, and t pk Represents the expected output data corresponding to the kth neuron in the output layer, y pk represents the actual output data corresponding to the kth neuron in the output layer, j = 1, 2, .., J, J represents the total number of neurons in the hidden layer, h pj represents the output of the jth neuron in the hidden layer.

[0025] Furthermore, the weight of the distributed cache preheating loading model is updated, including:

[0026] Determine the weight update amount as:

[0027]

[0028] Among them, Δw jk represents the weight update between the jth neuron in the hidden layer and the kth neuron in the output layer, Δw ij represents the weight update between the i-th neuron in the input layer and the j-th neuron in the hidden layer, i = 1, 2, .., I, I represents the total number of neurons in the input layer, η represents the learning rate, γ represents the adjustment coefficient, exp represents the exponential function with the natural constant e as the base, w jk represents the weight between the jth neuron in the hidden layer and the kth neuron in the output layer;

[0029] According to the weight update amount, the updated weight is determined as:

[0030] w jk '=w jk +Δw jk

[0031] w ij '=w ij +Δw ij

[0032] Among them, w jk represents the weight between the jth neuron in the hidden layer and the kth neuron in the output layer before updating, w jk ' represents the weight between the jth neuron in the hidden layer and the kth neuron in the output layer after the update, w ij represents the weight between the i-th neuron in the input layer and the j-th neuron in the hidden layer before updating, w ij ' represents the weight between the i-th neuron in the input layer and the j-th neuron in the hidden layer after the update, and L represents the current number of training times.

[0033] Furthermore, the learning rate η is:

[0034]

[0035]

[0036] Among them, t represents the gradient, n represents the learning rate calculation coefficient, represents the symbol for partial derivative, and w represents w ij or jk .

[0037] Furthermore, real-time user behavior data is collected and analyzed through a distributed cache preheating loading model with prediction capability to determine the behavior analysis results, including:

[0038] Collecting real-time user behavior data; the real-time user behavior data includes N user behaviors before the current moment;

[0039] Real-time user behavior data is used to construct input data of a distributed cache preheating loading model, and the input data is transmitted to the distributed cache preheating loading model to obtain a behavior analysis result.

[0040] Furthermore, based on the behavior analysis result, data in the distributed system is cached, including:

[0041] Based on the behavior analysis result, hot spot data corresponding to the analysis result is determined from the distributed system, and the hot spot data is cached.

[0042] The present invention provides a distributed cache preheating loading method based on machine learning, which can analyze the hot data of the system in advance, identify the hot data during system initialization and operation, and perform preload in advance; in the distributed cache preheating loading method, machine learning is introduced to identify the hot data of the system based on user behavior and system logs, and actively load the hot cache data, thereby further improving system performance and improving user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.

[0044] Figure 1 A flowchart of a distributed cache preheating loading method based on machine learning is provided in an embodiment of the present invention.

[0045] The above drawings have shown clear embodiments of the present invention, which will be described in more detail below. These drawings and text descriptions are not intended to limit the scope of the present invention in any way, but to illustrate the concept of the present invention for those skilled in the art by referring to specific embodiments. DETAILED DESCRIPTION

[0046] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present invention. Instead, they are merely examples of devices and methods consistent with some aspects of the present invention as detailed in the appended claims.

[0047] The embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0048] like Figure 1 As shown, an embodiment of the present invention provides a distributed cache preheating loading method based on machine learning, comprising:

[0049] S101, using system log data, wherein the system log data is historical data stored in a distributed system.

[0050] The use of system log data is a common practice in information technology and system management, which involves collecting, storing and analyzing log files generated by operating systems, applications and services. These logs record various events in a computer system or network, including but not limited to user activities, error messages, performance indicators and security events. Since the embodiments of the present invention mainly involve user behavior prediction, the corresponding data is cached to improve system performance, so the system log data can include historical user behavior data.

[0051] S102: Analyze the system log data to determine historical user behavior data, wherein the historical user behavior data includes user behavior and user behavior time.

[0052] The user behavior time mainly includes the month, the day of the month and the time of the day. The user behavior and the user behavior time can be formed into a tuple. In the subsequent use process, multiple tuples can be formed into a vector to realize data recognition.

[0053] S103: Use a machine learning algorithm to build a distributed cache preheating loading model, and use the distributed cache preheating loading model to learn historical user behavior data to obtain a distributed cache preheating loading model with predictive capabilities.

[0054] Using machine learning algorithms to build a distributed cache preheating loading model is a method that analyzes historical user behavior data, learns user access patterns and data usage trends, predicts data that may be accessed in the future, and loads this data into the cache in advance. Such a model can handle cache refresh and loading strategies more intelligently, improve cache utilization, and reduce data cold start time.

[0055] Optionally, you can clean and preprocess the data to remove outliers, fill in missing data, and transform the data into a format suitable for machine learning algorithms.

[0056] Machine learning algorithms can be set as time series prediction models (such as ARIMA), regression models, clustering algorithms (such as K-means), neural networks, and so on.

[0057] After obtaining a distributed cache pre-load model with predictive capabilities, you can also evaluate the model performance through methods such as cross-validation. Deploy the trained model to the production environment and adjust and optimize the model based on real-time user behavior data. Regularly evaluate the accuracy of model predictions and adjust model parameters or select new algorithms based on the evaluation results. Through continuous iteration and optimization, the model can better adapt to dynamically changing user behaviors, thereby more effectively improving the effect of distributed caching.

[0058] S104: Collect real-time user behavior data, and analyze the real-time user behavior data through a distributed cache preheating loading model with prediction capabilities to determine a behavior analysis result.

[0059] After learning the historical user behavior data through the distributed cache preheating loading model, the distributed cache preheating loading model can identify the data, thereby identifying the hot data and obtaining the behavior analysis results.

[0060] S105. Based on the behavior analysis results, cache the data in the distributed system to complete the distributed cache preheating loading based on machine learning.

[0061] The present invention provides a distributed cache preheating loading method based on machine learning, which can analyze the hot data of the system in advance, identify the hot data during system initialization and operation, and preload in advance. In the distributed cache preheating loading method, machine learning is introduced to identify the hot data of the system based on user behavior and system logs, and actively load the hot cache data, thereby further improving system performance and user experience.

[0062] In an embodiment of the present invention, using system log data includes: connecting to a distributed system through an API interface, and acquiring system log data stored in the distributed system.

[0063] In an embodiment of the present invention, a distributed cache preheating loading model is constructed using a machine learning algorithm, and historical user behavior data is learned through the distributed cache preheating loading model to obtain a distributed cache preheating loading model with predictive capabilities, including:

[0064] A distributed cache preheating loading model is built using machine learning algorithms.

[0065] Optionally, the embodiment of the present invention preferably adopts BP network as the distributed cache preheating loading model, but it is worth noting that this implementation is only an example, and other neural network models can also be used to construct the distributed cache preheating loading model.

[0066] BP (Back Propagation) neural network is a multi-layer feedforward neural network trained by the error back propagation algorithm. This network usually includes an input layer, one or more hidden layers, and an output layer. Each neuron is connected to each neuron in the next layer, and each connection has a corresponding weight. BP neural network automatically extracts features by learning the mapping relationship between input data and expected output, and is used for various machine learning tasks such as classification and regression.

[0067] Based on the historical user behavior data, N historical user behavior data at a historical time point are determined as sample data, and the N+1th historical user behavior data is determined as expected data.

[0068] Optionally, the N+2 and N+3 historical user behavior data can be used as expected data, and then multiple distributed cache preheating loading models can be trained respectively to predict hot data at multiple time points in the future, thereby improving system performance and user experience. It is worth noting that the above short-term behavior prediction is only an example, and the behavior at other time points can also be predicted to provide users with a better data experience.

[0069] The distributed cache preheating loading model is used to learn the association relationship between the sample data and the expected data, thereby obtaining a distributed cache preheating loading model with prediction capability.

[0070] Learning the association relationship between the sample data and the expected data through the distributed cache preheating loading model may include: training the distributed cache preheating loading model based on the association relationship between the sample data and the expected data and through an intelligent optimization algorithm (such as a particle swarm algorithm, a genetic algorithm, etc.).

[0071] In an embodiment of the present invention, the distributed cache preheating loading model is used to learn the association relationship between the sample data and the expected data to obtain a distributed cache preheating loading model with prediction capability, including:

[0072] Initialize the model parameters corresponding to the distributed cache preheating loading model, the model parameters are weight parameters. It is worth noting that threshold parameters may also be included, and the update method may be the same as the weight parameter.

[0073] Initializing the model parameters corresponding to the distributed cache preheating loading model may include: randomly initializing between an upper weight limit and a lower weight limit.

[0074] The sample data is used as input of the distributed cache preheating loading model to obtain actual output data of the distributed cache preheating loading model.

[0075] According to the expected data and the actual output data, the error function value corresponding to the distributed cache preheating loading model is obtained.

[0076] Optionally, in a BP neural network, an error function is usually used to measure the difference between the network output and the actual label. The commonly used error function is the mean square error (MSE).

[0077] It is determined whether the error function value is less than a preset threshold. If so, a distributed cache preheating loading model with prediction capability is obtained. Otherwise, the weight of the distributed cache preheating loading model is updated and the next training is started.

[0078] Optionally, you can also set a maximum number of training times. When the maximum number of training times is reached, you can stop training and obtain a distributed cache preheating loading model with predictive capabilities.

[0079] In an embodiment of the present invention, initializing the model parameters corresponding to the distributed cache preheating loading model includes:

[0080] Between the upper and lower limits of the parameters of the distributed cache preheating loading model, a random initialization method or a chaotic mapping initialization method is used to generate model parameters to achieve parameter initialization.

[0081] When the initialization method is adopted, the model parameters can be directly optimized using the parameter optimization method provided in the embodiment of the present invention. When the chaotic mapping initialization method is used, multiple parameter individuals can be generated in the solution space, and each parameter individual is equivalent to containing all the model parameters. Then, the parameter optimization method provided in the embodiment of the present invention is executed on different parameter individuals, which can effectively improve the algorithm search efficiency.

[0082] In deep learning, the initialization of model parameters is an important step, which has a great impact on the training process and final performance of the model. Common initialization methods include random initialization and initialization using specific algorithms (such as chaos mapping). Chaotic mapping is a mathematical method based on chaos theory, which involves mapping an initial value to another value through a nonlinear transformation. Chaotic mapping can make the generated parameter individuals more evenly distributed in the solution space, thereby improving the algorithm's search ability.

[0083] Although the mean square error can be used as the error function, the mean square error is less effective and may cause training deviations. Therefore, an embodiment of the present invention provides an error function that takes into account the saturation of neurons in the hidden layer, avoids the problem of local minimum of the loss function value, and can achieve better results.

[0084] In the embodiment of the present invention, the error function value is obtained by an error function, and the error function is:

[0085]

[0086] Where L represents the error function, p = 1, 2, .., P, P represents the total number of input data, k = 1, 2, .., K, K represents the total number of neurons in the output layer, and t pk Represents the expected output data corresponding to the kth neuron in the output layer, y pk represents the actual output data corresponding to the kth neuron in the output layer, j = 1, 2, .., J, J represents the total number of neurons in the hidden layer, h pj represents the output of the jth neuron in the hidden layer.

[0087] In an embodiment of the present invention, updating the weight of the distributed cache preheating loading model includes:

[0088] Determine the weight update amount as:

[0089]

[0090] Among them, Δw jk represents the weight update between the jth neuron in the hidden layer and the kth neuron in the output layer, Δw ij represents the weight update between the i-th neuron in the input layer and the j-th neuron in the hidden layer, i = 1, 2, .., I, I represents the total number of neurons in the input layer, η represents the learning rate, γ represents the adjustment coefficient, exp represents the exponential function with the natural constant e as the base, w jk represents the weight between the jth neuron in the hidden layer and the kth neuron in the output layer.

[0091] According to the weight update amount, the updated weight is determined as:

[0092] w jk '=w jk +Δw jk

[0093] w ij '=w ij +Δw ij

[0094] Among them, w jk represents the weight between the jth neuron in the hidden layer and the kth neuron in the output layer before updating, w jk ' represents the weight between the jth neuron in the hidden layer and the kth neuron in the output layer after the update, w ij represents the weight between the i-th neuron in the input layer and the j-th neuron in the hidden layer before updating, w ij ' represents the weight between the i-th neuron in the input layer and the j-th neuron in the hidden layer after the update, and L represents the current number of training times.

[0095] In this embodiment of the present invention, the learning rate η is:

[0096]

[0097]

[0098] Among them, t represents the gradient, n represents the learning rate calculation coefficient, represents the symbol for partial derivative, and w represents w ij or jk .

[0099] By setting a variable learning rate, we can effectively balance the algorithm's early global search and later fine search, thereby increasing the algorithm's effectiveness.

[0100] Optionally, the learning rate η may also adopt a fixed value, for example, 0.01 or other values, which will not be described here.

[0101] In an embodiment of the present invention, real-time user behavior data is collected, and the real-time user behavior data is analyzed by a distributed cache preheating loading model with prediction capability to determine the behavior analysis result, including:

[0102] Collect real-time user behavior data, which includes N user behaviors before the current moment.

[0103] Real-time user behavior data is used to construct input data of a distributed cache preheating loading model, and the input data is transmitted to the distributed cache preheating loading model to obtain a behavior analysis result.

[0104] In an embodiment of the present invention, based on the behavior analysis result, data in a distributed system is cached, including:

[0105] Based on the behavior analysis result, hot spot data corresponding to the analysis result is determined from the distributed system, and the hot spot data is cached.

[0106] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes. The schemes in the embodiments of the present invention may be implemented in various computer languages, for example, object-oriented programming language Java and literal scripting language JavaScript, etc.

[0107] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0108] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0109] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0110] Although the preferred embodiments of the present invention have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0111] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A distributed cache preheating loading method based on machine learning, characterized in that: include: Using system log data; wherein the system log data is historical data stored in a distributed system; Parsing the system log data to determine historical user behavior data; wherein the historical user behavior data includes user behavior and user behavior time; A distributed cache preheating loading model is constructed using a machine learning algorithm, and historical user behavior data is learned through the distributed cache preheating loading model to obtain a distributed cache preheating loading model with predictive capabilities; Collect real-time user behavior data, and analyze the real-time user behavior data through a distributed cache preheating loading model with prediction capabilities to determine the behavior analysis results; Based on the behavior analysis results, the data in the distributed system is cached to complete the distributed cache preheating loading based on machine learning.

2. The distributed cache preheating loading method based on machine learning according to claim 1 is characterized in that: Using system log data, including: connecting to the distributed system through the API interface, and obtaining the system log data stored in the distributed system.

3. The distributed cache preheating loading method based on machine learning according to claim 1 is characterized in that: A distributed cache preheating loading model is constructed using a machine learning algorithm. The distributed cache preheating loading model is used to learn historical user behavior data to obtain a distributed cache preheating loading model with predictive capabilities, including: Use machine learning algorithms to build a distributed cache preheating loading model; Based on the historical user behavior data, determine N historical user behavior data at a historical time point as sample data, and determine the N+1th historical user behavior data as expected data; The distributed cache preheating loading model is used to learn the association relationship between the sample data and the expected data, thereby obtaining a distributed cache preheating loading model with prediction capability.

4. The distributed cache preheating loading method based on machine learning according to claim 3 is characterized in that: The distributed cache preheating loading model is used to learn the association between the sample data and the expected data, so as to obtain a distributed cache preheating loading model with prediction capability, including: Initialize the model parameters corresponding to the distributed cache preheating loading model, where the model parameters are weight parameters; Using the sample data as input of a distributed cache preheating loading model, obtaining actual output data of the distributed cache preheating loading model; According to the expected data and the actual output data, an error function value corresponding to the distributed cache preheating loading model is obtained; It is determined whether the error function value is less than a preset threshold. If so, a distributed cache preheating loading model with prediction capability is obtained. Otherwise, the weight of the distributed cache preheating loading model is updated and the next training is started.

5. The distributed cache preheating loading method based on machine learning according to claim 4 is characterized in that: Initialize the model parameters corresponding to the distributed cache preheat loading model, including: Between the upper and lower limits of the parameters of the distributed cache preheating loading model, a random initialization method or a chaotic mapping initialization method is used to generate model parameters to achieve parameter initialization.

6. The distributed cache preheating loading method based on machine learning according to claim 5 is characterized in that: The error function value is obtained by an error function, and the error function is: Where L represents the error function, p = 1, 2, .., P, P represents the total number of input data, k = 1, 2, .., K, K represents the total number of neurons in the output layer, and t pk Represents the expected output data corresponding to the kth neuron in the output layer, y pk represents the actual output data corresponding to the kth neuron in the output layer, j = 1, 2, .., J, J represents the total number of neurons in the hidden layer, h pj represents the output of the jth neuron in the hidden layer.

7. The distributed cache preheating loading method based on machine learning according to claim 6 is characterized in that: Update the weights of the distributed cache preheat loading model, including: Determine the weight update amount as: Among them, Δw jk represents the weight update between the jth neuron in the hidden layer and the kth neuron in the output layer, Δw ij represents the weight update between the i-th neuron in the input layer and the j-th neuron in the hidden layer, i = 1, 2, .., I, I represents the total number of neurons in the input layer, η represents the learning rate, γ represents the adjustment coefficient, exp represents the exponential function with the natural constant e as the base, w jk represents the weight between the jth neuron in the hidden layer and the kth neuron in the output layer; According to the weight update amount, the updated weight is determined as: In jk '=in jk +Δw jk In ij '=in ij +Δw ij Among them, w jk represents the weight between the jth neuron in the hidden layer and the kth neuron in the output layer before updating, w jk ' represents the weight between the jth neuron in the hidden layer and the kth neuron in the output layer after the update, w ij represents the weight between the i-th neuron in the input layer and the j-th neuron in the hidden layer before updating, w ij ' represents the weight between the i-th neuron in the input layer and the j-th neuron in the hidden layer after the update, and L represents the current number of training times.

8. The distributed cache preheating loading method based on machine learning according to claim 7 is characterized in that: The learning rate η is: Among them, t represents the gradient, n represents the learning rate calculation coefficient, represents the symbol for partial derivative, and w represents w ij or jk .

9. The distributed cache preheating loading method based on machine learning according to claim 1 is characterized in that: Collect real-time user behavior data, and analyze the real-time user behavior data through a distributed cache preheating loading model with prediction capabilities to determine the behavior analysis results, including: Collecting real-time user behavior data; the real-time user behavior data includes N user behaviors before the current moment; Real-time user behavior data is used to construct input data of a distributed cache preheating loading model, and the input data is transmitted to the distributed cache preheating loading model to obtain a behavior analysis result.

10. The distributed cache preheating loading method based on machine learning according to claim 9 is characterized in that: Based on the behavior analysis result, data in the distributed system is cached, including: Based on the behavior analysis result, hot spot data corresponding to the analysis result is determined from the distributed system, and the hot spot data is cached.

Images