Easily accessible prediction method and device, electronic equipment and storage medium

Abstract

The invention provides an easy-to-visit prediction method and device, electronic equipment and a storage medium, and the method comprises the steps: obtaining first service quality data and first behavior data related to an access behavior of a user in a first time period, inputting the first service quality data and the first behavior data into a pre-trained easy-to-visit prediction model, obtaining an easy-to-visit prediction result, output by an easy-to-visit prediction model, of the predicted user in a second time period, the easy-to-visit prediction model comprising a target prediction model and a target classification model, and the target prediction model being used for predicting second service quality data and second behavior data of the user in the second time period, the target classification model is used for determining an easy-to-visit prediction result according to the second service quality data and the second behavior data, so that complex modes in the user characteristics and the behavior data can be deeply analyzed and captured, accurate prediction of the user access tendency is realized, and the reliability and the accuracy of the easy-to-visit prediction result can be effectively improved.

Description

Technical Field

[0001] The present disclosure relates to the field of artificial intelligence technology, and in particular to an accessibility prediction method, device, electronic device, and storage medium. Background Art

[0002] Related technologies for accessibility prediction primarily rely on traditional statistical learning methods or rudimentary machine learning techniques. These approaches exhibit significant limitations when processing complex user behavior data and analyzing multidimensional features. This is particularly true when dealing with large-scale, high-dimensional, and unbalanced datasets. These methods fail to fully utilize all available data and struggle to capture subtle changes and complex patterns in user behavior. This results in models being unable to effectively predict future user visitation trends, impacting the accuracy and reliability of accessibility predictions. Summary of the Invention

[0003] The present disclosure proposes an accessibility prediction method, apparatus, electronic device, storage medium, and computer program product, aiming to solve the technical problems in the related art at least to a certain extent.

[0004] An embodiment of a first aspect of the present disclosure proposes an accessibility prediction method, including: obtaining first service quality data and first behavior data related to a user's access behavior in a first time period; inputting the first service quality data and the first behavior data into a pre-trained accessibility prediction model to obtain an accessibility prediction result of the user in a second time period output by the accessibility prediction model, wherein the accessibility prediction model includes: a target prediction model and a target classification model, the target prediction model is used to predict the second service quality data and second behavior data of the user in the second time period, and the target classification model is used to determine the accessibility prediction result based on the second service quality data and the second behavior data, and the first time period is before the second time period.

[0005] An embodiment of a second aspect of the present disclosure proposes an accessibility prediction device, including: an acquisition module, used to obtain first service quality data and first behavior data related to the access behavior of a user in a first time period; a processing module, used to input the first service quality data and the first behavior data into a pre-trained accessibility prediction model to obtain an accessibility prediction result of the user in a second time period output by the accessibility prediction model, wherein the accessibility prediction model includes: a target prediction model and a target classification model, the target prediction model is used to predict the second service quality data and second behavior data of the user in the second time period, and the target classification model is used to determine the accessibility prediction result based on the second service quality data and the second behavior data, and the first time period is before the second time period.

[0006] A third aspect of the present disclosure provides an electronic device, comprising: a processor; and a memory for storing processor-executable instructions; wherein the processor is configured to execute the instructions to implement an accessibility prediction method.

[0007] A fourth aspect of the present disclosure provides a computer-readable storage medium. When instructions in the computer-readable storage medium are executed by a processor of an electronic device, the electronic device is enabled to perform an accessibility prediction method.

[0008] A fifth aspect of the present disclosure provides a computer program product, including a computer program, wherein the computer program is executed by a processor to perform an accessibility prediction method.

[0009] The accessibility prediction method, apparatus, electronic device, storage medium, and computer program product proposed in this embodiment have at least the following beneficial effects: obtaining first quality of service data and first behavior data related to a user's access behavior in a first time period, inputting the first quality of service data and first behavior data into a pre-trained accessibility prediction model to obtain an accessibility prediction result output by the accessibility prediction model for predicting the user's accessibility in a second time period, wherein the accessibility prediction model includes: a target prediction model and a target classification model, the target prediction model is used to predict the user's second quality of service data and second behavior data in the second time period, and the target classification model is used to determine the accessibility prediction result based on the second quality of service data and second behavior data, with the first time period before the second time period. Therefore, it is possible to deeply analyze and capture complex patterns in user characteristics and behavior data, achieve accurate prediction of user access tendencies, and thus effectively improve the reliability and accuracy of the accessibility prediction result.

[0010] Additional aspects and advantages of the present disclosure will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] The above and / or additional aspects and advantages of the present disclosure will become apparent and readily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0012] Figure 1 is a flowchart of an accessibility prediction method according to the first embodiment of the present disclosure;

[0013] Figure 2 is a flow chart of an accessibility prediction method according to the second embodiment of the present disclosure;

[0014] Figure 3 is a schematic diagram of a calculation process of a Triformer model according to an embodiment of the present disclosure;

[0015] Figure 4 is a flow chart of an accessibility prediction method according to the third embodiment of the present disclosure;

[0016] Figure 5 is a block diagram of an accessibility prediction device according to the present disclosure;

[0017] Figure 6 A block diagram of an exemplary electronic device suitable for implementing embodiments of the present disclosure is shown. DETAILED DESCRIPTION

[0018] The following describes in detail embodiments of the present disclosure, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended only to explain the present disclosure and are not to be construed as limiting the present disclosure. On the contrary, the embodiments of the present disclosure include all variations, modifications, and equivalents that fall within the spirit and scope of the appended claims.

[0019] It should be noted that the execution subject of the accessibility prediction method of this embodiment may be an accessibility prediction device, which may be implemented by software and / or hardware. The device may be configured in an electronic device, which may include but is not limited to a terminal, a server, etc.

[0020] It should be noted that the processes of acquiring, storing, using, and processing information in the disclosed technical solution comply with the relevant provisions of national laws and regulations and do not violate public order and good morals.

[0021] Figure 1 is a flow chart of the accessibility prediction method according to the first embodiment of the present disclosure, as shown in FIG. Figure 1 As shown, the method includes:

[0022] S101: Acquire first service quality data and first behavior data related to the access behavior of a user in a first time period.

[0023] The user may be, for example, a user of a certain platform or a certain service.

[0024] Accessibility prediction involves comprehensively analyzing historical user behavior data, real-time feature data, and external factors to predict and assess user accessibility, service experience quality, and potential issues over a specific time period. Its core goal is to scientifically identify potential access barriers or opportunities for a superior experience, thereby providing decision support for service providers, optimizing resource allocation, and improving user satisfaction.

[0025] For example, accessibility prediction can be to obtain the user's usage data of a certain platform in a historical period, and predict based on the data whether the user belongs to the "accessible category" (high-quality experience) or the "non-accessible category" (potential problems) in a certain period of time in the future, so as to provide personalized services (such as priority resource protection) for accessible users and provide early warnings and solve problems for non-accessible users, thereby effectively improving user satisfaction.

[0026] The first time period refers to a historical time period, and the first time period can be set adaptively without any restriction.

[0027] The user's access behavior to the platform or service within the first time period may be, for example, browsing behavior, usage behavior, etc., and there is no restriction on this.

[0028] Among them, the data used to describe the service quality provided by the platform or service to the user is the first service quality data. The first service quality data can be, for example: quality problems encountered by users during web browsing, voice communication, and video watching (respectively marked as poor Internet quality, poor voice quality, and poor video quality), etc., and there is no restriction on this.

[0029] Among them, the data used to describe the user's access behavior is the first behavior data, and the first behavior data can be, for example: access frequency data, consumption amount data, likes data, etc., without limitation.

[0030] That is, in the embodiment of the present disclosure, the initial service quality data and initial behavior data related to the user's access behavior in the first time period may be obtained, and the aforementioned initial service quality data and initial behavior data may be value-filled and box-typed. Figure 4 The quantile detection method completes the filling of missing values and the identification and processing of outliers to obtain the first service quality data and the first behavior data to ensure the quality and consistency of the data.

[0031] S102: Inputting the first service quality data and the first behavior data into a pre-trained accessibility prediction model to obtain an accessibility prediction result of the user in a second time period output by the accessibility prediction model.

[0032] Among them, the accessibility prediction model includes: target prediction model and target classification model.

[0033] In the embodiment of the present disclosure, the target prediction model is a triangular variable-specific attention for long sequence multivariate time series forecasting model (Triformer).

[0034] In the embodiment of the present disclosure, the target classification model is an integrated classification model based on random under-sampling (Random Under-Sampling Boosting, RUSboost).

[0035] That is to say, in the embodiment of the present disclosure, the initial prediction model and the initial classification model may be iteratively trained respectively until the initial prediction model and the initial classification model converge to obtain the target prediction model and the target classification model.

[0036] The target prediction model is used to predict second service quality data and second behavior data of the user in a second time period, and the target classification model is used to determine an accessibility prediction result based on the second service quality data and the second behavior data.

[0037] The first time period is before the second time period.

[0038] In combination with the above examples, the user's accessibility prediction result may refer to: "accessible category" (high-quality experience) or "non-accessible category" (potential problems), without limitation.

[0039] That is to say, in the embodiment of the present disclosure, after obtaining the first service quality data and first behavior data related to the user's access behavior in the first time period, the first service quality data and the first behavior data can be input into a pre-trained accessibility prediction model to obtain the accessibility prediction result of the user in the second time period output by the accessibility prediction model.

[0040] In an embodiment of the present disclosure, first service quality data and first behavior data related to the user's access behavior in a first time period are obtained, and the first service quality data and the first behavior data are input into a pre-trained accessibility prediction model to obtain an accessibility prediction result output by the accessibility prediction model for predicting the user's accessibility in a second time period. The accessibility prediction model includes: a target prediction model and a target classification model. The target prediction model is used to predict the user's second service quality data and second behavior data in the second time period. The target classification model is used to determine the accessibility prediction result based on the second service quality data and the second behavior data. The first time period is before the second time period. Therefore, it is possible to deeply analyze and capture complex patterns in user characteristics and behavior data, achieve accurate prediction of user access tendencies, and thus effectively improve the reliability and accuracy of the accessibility prediction result.

[0041] Figure 2 is a flow chart of the accessibility prediction method according to the second embodiment of the present disclosure, as shown in FIG. Figure 2 As shown, the method includes:

[0042] S201: Acquire first service quality data and first behavior data related to the access behavior of a user in a first time period.

[0043] The detailed description of S201 can be found in the above embodiment and will not be repeated here.

[0044] S202: Input the first service quality data and the first behavior data into a target prediction model to obtain target prediction data output by the target prediction model, wherein the target prediction data includes: second service quality data and second behavior data.

[0045] In an embodiment of the present disclosure, after obtaining the first service quality data and first behavior data related to the user's access behavior in the first time period, the first service quality data and the first behavior data can be input into the target prediction model to obtain the target prediction data output by the target prediction model.

[0046] In the embodiments of this disclosure, see Figure 3 , Figure 3 The calculation flow diagram of the Triformer model according to an embodiment of the present disclosure is shown, that is, after the first service quality data and the first behavior data are input into the target prediction model, the target prediction model will be based on Figure 3 The processing flow shown processes the first quality of service data and the first behavior data to obtain target prediction data.

[0047] S203: Input the target prediction data into the target classification model to obtain the accessibility prediction result output by the target classification model.

[0048] In an embodiment of the present disclosure, after inputting the first service quality data and the first behavior data into the target prediction model to obtain the target prediction data output by the target prediction model, the target prediction data can be input into the target classification model to obtain the accessibility prediction result output by the target classification model.

[0049] In the embodiment of the present disclosure, the target prediction model is obtained by optimizing the initial parameters of the ensemble classification model based on random undersampling based on the differential evolution algorithm to determine the target model parameters, and updating the initial parameters of the ensemble classification model based on random undersampling according to the target model parameters.

[0050] Differential evolution (DE) is an evolutionary algorithm used to solve real-valued optimization problems. It iteratively searches for the global optimal solution. In each iteration, DE optimizes the individuals in the population through mutation, crossover, and selection operations, gradually approaching the global optimal solution.

[0051] RUSboost is an ensemble learning algorithm for classification problems, particularly suitable for scenarios with imbalanced data. It constructs a series of base classifiers by randomly downsampling samples of the majority class and then using the AdaBoost strategy to gradually increase the weight of misclassified samples.

[0052] In the disclosed embodiments, by combining the DE algorithm with RUSboost, we can optimize RUSboost parameters to improve its classification performance on imbalanced data. The DE algorithm's global search capability can help us find the optimal RUSboost parameter settings, including weight update rules, downsampling ratios, etc., without any restrictions.

[0053] Among them, the regular training process of RUSboost includes:

[0054] (1) Hyperparameter initialization, the specific process is:

[0055]

[0056] Where N is the total number of training samples.

[0057] (2) Iteratively train the base classifier. The specific process is as follows:

[0058] For each iteration from m=1 to M, random downsampling is performed to sample the majority class samples to balance the data; using the current distribution D m Train the base classifier h m ; Calculate h m The error rate ∈ on the training data m; Calculate h m Weight:

[0059]

[0060] Update sample weight D m+1 (i).

[0061] (3) The model output is:

[0062]

[0063] As can be seen from the conventional training process described above, RUSBoost optimizes the handling of imbalanced binary classification problems by combining random undersampling and AdaBoost. This approach primarily increases the model's focus on the minority class by randomly undersampling majority class examples in each iteration. While this approach is effective in improving the recognition rate of the minority class, it has some drawbacks, such as potential information loss, sensitivity to initial parameter selection, and difficulty adjusting parameters on extremely imbalanced datasets, which can affect the model's generalization and stability.

[0064] Therefore, in the embodiment of the present disclosure, an effective way is provided by introducing the differential evolution (DE) algorithm for parameter optimization. The global search capability of DE enables it to find optimized solutions in a wide range of parameter spaces, reduce dependence on initial parameter settings, and improve the stability and generalization ability of the model. In addition, the robustness and easy implementation of the DE algorithm make it efficient and adaptable when dealing with complex optimization problems. By using DE to optimize the parameters of RUSBoost, the needs of handling unbalanced binary classification problems can be more accurately balanced, effectively improving the prediction accuracy of the minority class while maintaining the optimization of the overall model performance.

[0065] The DE optimization process of RUSboost includes:

[0066] (1)DE optimization initialization

[0067] g=x1,x2,…,x NP ;

[0068] Among them, x i is the i-th candidate solution of the RUSboost parameter, and NP is the population size.

[0069] (2) Mutation operation

[0070] v i,G =x r1,G +F·(x r2,G -x r3,G );

[0071] Where G is the current generation, r1, r2, r3 are randomly selected and different index numbers, and F is the differential weight.

[0072] (3) Crossover operation, i.e. generating the test vector u i ,G, the process can be expressed as:

[0073]

[0074] Among them, CR is the crossover probability, rand j is a random number between [0,1], j rand is a randomly selected index that ensures that at least one variable comes from the mutation vector.

[0075] (4) Selection operation, that is, selecting individuals with higher fitness as candidate solutions for the next generation:

[0076]

[0077] Where f(·) is the fitness function, which is usually the performance indicator of the classifier on the validation set.

[0078] In the embodiment of the present disclosure, after optimizing the initial parameters of the ensemble classification model based on random undersampling based on the differential evolution algorithm to determine the target model parameters, and updating the initial parameters of the ensemble classification model based on random undersampling according to the target model parameters to obtain the target classification model, the target prediction data can be input into the target classification model to obtain the accessibility prediction result output by the target classification model. Thus, the global search capability of the DE algorithm can be used to find the optimal RUSboost parameter setting, optimize the parameters of RUSboost, so as to improve its classification performance on imbalanced data, effectively solve the data imbalance problem, and then when the target prediction data is processed based on the target classification model, the classification prediction accuracy can be effectively improved.

[0079] In an embodiment of the present disclosure, first service quality data and first behavior data related to the user's access behavior in a first time period are obtained, and then the first service quality data and the first behavior data are input into a target prediction model to obtain target prediction data output by the target prediction model, wherein the target prediction data includes: second service quality data and second behavior data, and the target prediction data is input into a target classification model to obtain the accessibility prediction result output by the target classification model. Thus, the target prediction model can be combined to accurately predict the user's target prediction data in the second time period, and then the target classification model can realize accurate prediction of the user's accessibility based on the target prediction data.

[0080] Figure 4is a flow chart of an accessibility prediction method according to the third embodiment of the present disclosure, as shown in FIG. Figure 4 As shown, the method includes:

[0081] S401: Acquire first service quality data and first behavior data related to the access behavior of a user in a first time period.

[0082] The detailed description of S401 can be found in the above embodiment and will not be repeated here.

[0083] S402: Generate a target data sequence according to the first quality of service data and the first behavior data.

[0084] In an embodiment of the present disclosure, after obtaining the first service quality data and first behavior data related to the user's access behavior within the first time period, the first service quality data and the first behavior data can be sorted according to the timestamp information corresponding to the first service data and the first behavior data to obtain a target data sequence that is continuous in time.

[0085] For example, in the specific scenario of Yifang prediction, using X1, X2, X3, ..., X n To represent the user's feature sequence, where each X i Represents a specific feature dimension. For example, X1 is the category of poor Internet quality (features such as the proportion of poor 4G and 5G Internet quality), X2 is the category of poor voice quality (features such as the number of voice dissatisfaction and RSRP sampling points), and so on. Specifically, let t be the time step, is the value of the i-th feature at time t. For a time series data with a history length of T, we can express the target data sequence as:

[0086]

[0087] Where t∈{1,2,…,T}, X t Represents the set of all features at time t.

[0088] S403: Divide the target data sequence based on a preset time step to obtain a plurality of initial data blocks.

[0089] The preset time step can be combined with the accessibility prediction requirements in actual business scenarios and set adaptively without any restrictions.

[0090] In the embodiment of the present disclosure, the target data sequence may be divided based on a preset time step S, and the initial data blocks obtained by the division may be expressed as:

[0091] X p =(X (p-1)S+1 ,X (p-1)S+2 ,…,XpS );

[0092] Among them, X p Represents the Pth patch, containing data from (p-1)S+1 to pS time steps.

[0093] S404: Input each initial data block into the target prediction model to obtain initial prediction data output by the target prediction model corresponding to each initial data block.

[0094] In an embodiment of the present disclosure, after the target data sequence is divided based on a preset time step to obtain multiple initial data blocks, each initial data block can be input into the target prediction model respectively to obtain initial prediction data output by the target prediction model corresponding to each initial data block.

[0095] Optionally, in some embodiments, each initial data block is input into the target prediction model respectively to obtain initial prediction data output by the target prediction model corresponding to each initial data block. Each target input data can be input into the i-th processing layer of the target prediction model to obtain i-th layer output data corresponding to each target input data output by the i-th processing layer, wherein i is a positive integer. When i=1, the target input data is the initial data block. When i>1, the target input data is the i-th layer output data. The initial prediction data is then generated based on the output data of each processing layer in the i processing layers in the target prediction model.

[0096] That is to say, in the embodiment of the present disclosure, see above Figure 3 , the initial data block can be input into the first processing layer of the target prediction model to obtain the first output data output by the first processing layer, and then the first output data is used as the input of the second processing layer of the target prediction model to obtain the second output data output by the second processing layer, and so on, until the output data of each layer of the target prediction model is obtained, and the initial prediction data is generated according to the output data of each layer of the target prediction model.

[0097] Optionally, in some embodiments, each target input data is input into the i-th processing layer of the target prediction model to obtain the i-th output data corresponding to each target input data output by the i-th processing layer. The target input data can be input into the i-th processing layer of the target prediction model to determine the first query matrix, the first key matrix and the first value matrix corresponding to the target input data, and the first attention weight matrix corresponding to the target input data is determined according to the first query matrix, the first key matrix and the first value matrix. Initial output data is generated according to the first attention weight and the first value matrix, and the initial output data is processed based on the nonlinear activation function and the state transition function to obtain the i-th output data.

[0098] In the embodiment of the present disclosure, each patchX p This is processed through the triangular structure of the Triformer model. At each level, an attention mechanism is applied to the patch to capture the long-term dependencies between features.

[0099] T p =Φ(A(T p ,X p ));

[0100] Among them, A(T p ,X p ) is a layer based on the output T of the previous layer p and the current patchx p The self-attention function computed. Φ represents the nonlinear function applied to the self-attention output.

[0101] In the disclosed embodiment, the self-attention mechanism generally includes three key components: query (Q), key (K), and value (V). p Every element X pi , calculate its query matrix (Q), key matrix (K), value matrix (V) mapping:

[0102] Q(X pi )=X pi W Q ;

[0103] K(X pi )=X pi W K ;

[0104] V(X pi )=X pi W V ;

[0105] Among them, W Q , W K and W V is a learning parameter.

[0106] In the disclosed embodiment, the attention score can be calculated using the dot product of the query and the key, and then the softmax function is applied to obtain the attention weight:

[0107]

[0108] Among them, d k is the dimension of the key vector, used to scale the dot product so that the gradient is stable.

[0109] In the embodiment of the present disclosure, after obtaining the initial output data, the initial output data can be processed based on the nonlinear activation function and the state transition function to obtain the i-th layer output data. The processing process can be expressed as:

[0110] T p+1 =g(Θ1T p +b1)⊙σ(Θ2T p +b2)+T p ;

[0111] Where g is the state transition function, σ is the nonlinear activation function, ⊙ represents element-wise multiplication, Θ1 and Θ2 are weight matrices, and b1 and b2 are bias vectors.

[0112] Optionally, in some embodiments, initial prediction data is generated based on the output data of each processing layer in the i processing layers in the target prediction model, and the output data of each processing layer in the i processing layers in the target prediction model can be fused to obtain the initial prediction data.

[0113] In the disclosed embodiment, through the last layer of the Triformer model, we obtain the prediction of the future value of each feature. The final output is ′ (Initial prediction data) combines the output data from each layer of the model to generate the initial prediction data prediction for the next time step:

[0114] O ′ =θ(T1 ′ ,…,T k ′ ,…,T p ′ );

[0115] Here, θ is a function that fuses the output data of different layers.

[0116] S405: performing fusion processing on the initial prediction data corresponding to each initial data block to obtain target prediction data.

[0117] In the embodiment of the present disclosure, each initial data block is input into the target prediction model respectively to obtain the initial prediction data output by the target prediction model corresponding to each initial data block. Then, the initial prediction data corresponding to each initial data block can be fused to obtain the target prediction data.

[0118] Specifically, in the embodiment of the present disclosure, after determining the initial prediction data corresponding to each initial data block, the initial prediction data corresponding to multiple initial data blocks can be weighted and summed based on a preset weight coefficient to obtain target prediction data, and then, the target prediction data can be processed based on the target classification model to obtain the accessibility prediction result.

[0119] S406: Input the target prediction data into the target classification model to obtain the accessibility prediction result output by the target classification model.

[0120] The specific description of S406 can be found in the above embodiment and is not limited here.

[0121] In an embodiment of the present disclosure, first service quality data and first behavior data related to the user's access behavior within a first time period are obtained, a target data sequence is generated based on the first service quality data and the first behavior data, the target data sequence is divided based on a preset time step to obtain a plurality of initial data blocks, each initial data block is input into a target prediction model to obtain initial prediction data output by the target prediction model corresponding to each initial data block, the initial prediction data corresponding to each initial data block are fused to obtain target prediction data, the target prediction data are input into a target classification model to obtain accessibility prediction results output by the target classification model, thereby enabling in-depth analysis and capture of complex patterns in user characteristics and behavior data, achieving accurate prediction of user access tendencies, and effectively improving the reliability and accuracy of accessibility prediction results.

[0122] Figure 5 is a block diagram of an accessibility prediction device according to the present disclosure, such as Figure 5 As shown, the accessibility prediction device 50 includes:

[0123] An acquisition module 501 is configured to acquire first service quality data and first behavior data related to an access behavior of a user within a first time period;

[0124] Processing module 502 is used to input the first service quality data and the first behavior data into a pre-trained accessibility prediction model to obtain an accessibility prediction result of the user in the second time period output by the accessibility prediction model, wherein the accessibility prediction model includes: a target prediction model and a target classification model, the target prediction model is used to predict the second service quality data and second behavior data of the user in the second time period, and the target classification model is used to determine the accessibility prediction result based on the second service quality data and the second behavior data, and the first time period is before the second time period.

[0125] In some embodiments of the present disclosure, the processing module 502 is further configured to:

[0126] Inputting the first service quality data and the first behavior data into a target prediction model to obtain target prediction data output by the target prediction model, wherein the target prediction data includes: second service quality data and second behavior data;

[0127] The target prediction data is input into the target classification model to obtain the accessibility prediction result output by the target classification model.

[0128] In some embodiments of the present disclosure, the processing module 502 is further configured to:

[0129] generating a target data sequence according to the first quality of service data and the first behavior data;

[0130] Dividing the target data sequence based on a preset time step to obtain multiple initial data blocks;

[0131] Input each initial data block into the target prediction model to obtain initial prediction data output by the target prediction model corresponding to each initial data block;

[0132] The initial prediction data corresponding to each initial data block are fused to obtain target prediction data.

[0133] In some embodiments of the present disclosure, the processing module 502 is further configured to:

[0134] Input each target input data into the i-th processing layer of the target prediction model to obtain the i-th layer output data corresponding to each target input data output by the i-th processing layer, where i is a positive integer. When i=1, the target input data is the initial data block; when i>1, the target input data is the i-th layer output data;

[0135] Generate initial prediction data based on the output data of each processing layer in the i processing layers in the target prediction model.

[0136] In some embodiments of the present disclosure, the processing module 502 is further configured to:

[0137] The output data of each processing layer in the target prediction model is fused to obtain the initial prediction data.

[0138] In some embodiments of the present disclosure, the processing module 502 is further configured to:

[0139] Inputting the target input data into the i-th processing layer of the target prediction model to determine a first query matrix, a first key matrix, and a first value matrix corresponding to the target input data;

[0140] Determining a first attention weight matrix corresponding to the target input data based on the first query matrix, the first key matrix, and the first value matrix;

[0141] Generate initial output data according to the first attention weight and the first value matrix;

[0142] The initial output data is processed based on the nonlinear activation function and the state transition function to obtain the i-th layer output data.

[0143] In some embodiments of the present disclosure, the target prediction model is obtained by optimizing the initial parameters of the ensemble classification model based on random undersampling based on the differential evolution algorithm to determine the target model parameters, and updating the initial parameters of the ensemble classification model based on random undersampling according to the target model parameters.

[0144] In some embodiments of the present disclosure, the target prediction model is a long sequence multivariate time series prediction model based on triangular variable specific attention.

[0145] In an embodiment of the present disclosure, first service quality data and first behavior data related to the user's access behavior in a first time period are obtained, and the first service quality data and the first behavior data are input into a pre-trained accessibility prediction model to obtain an accessibility prediction result output by the accessibility prediction model for predicting the user's accessibility in a second time period. The accessibility prediction model includes: a target prediction model and a target classification model. The target prediction model is used to predict the user's second service quality data and second behavior data in the second time period. The target classification model is used to determine the accessibility prediction result based on the second service quality data and the second behavior data. The first time period is before the second time period. Therefore, it is possible to deeply analyze and capture complex patterns in user characteristics and behavior data, achieve accurate prediction of user access tendencies, and thus effectively improve the reliability and accuracy of the accessibility prediction result.

[0146] In order to implement the above embodiments, the present application also proposes an electronic device, comprising: a processor, and a memory communicatively connected to the processor; the memory storing computer-executable instructions; and the processor executing the computer-executable instructions stored in the memory to implement the accessibility prediction method provided in the above embodiments.

[0147] In order to implement the above embodiments, the present application further proposes a computer-readable storage medium, in which computer-executable instructions are stored. When the computer-executable instructions are executed by a processor, they are used to implement the accessibility prediction method provided by the above embodiments.

[0148] Figure 6 A block diagram of an exemplary electronic device suitable for implementing embodiments of the present disclosure is shown.

[0149] Figure 6 The electronic device 6 shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present disclosure.

[0150] like Figure 6 As shown, electronic device 6 is implemented as a general purpose computing device. Components of electronic device 6 may include, but are not limited to, one or more processors or processing units 16, memory 28, and a bus 18 connecting various system components (including memory 28 and processing unit 16).

[0151] Bus 18 represents one or more of several types of bus structures, including a memory bus or memory controller, a peripheral bus, an accelerated graphics port, a processor, or a local bus using any of a variety of bus architectures. Examples of such architectures include, but are not limited to, the Industry Standard Architecture (ISA) bus, the Micro Channel Architecture (MAC) bus, the Enhanced ISA bus, the Video Electronics Standards Association (VESA) local bus, and the Peripheral Component Interconnection (PCI) bus.

[0152] The electronic device 6 typically includes a variety of computer system readable media. These media can be any available media that can be accessed by the electronic device 6, including volatile and non-volatile media, removable and non-removable media.

[0153] The memory 28 may include computer system readable media in the form of volatile memory, such as random access memory (RAM) 30 and / or cache memory 32. The electronic device 6 may further include other removable / non-removable, volatile / non-volatile computer system storage media. By way of example only, the storage system 34 may be used to read and write non-removable, non-volatile magnetic media ( Figure 6 Not shown, often called a "hard drive").

[0154] although Figure 6 Although not shown, a disk drive for reading and writing to a removable non-volatile disk (e.g., a "floppy disk"), and an optical disk drive for reading and writing to a removable non-volatile optical disk (e.g., a Compact Disc Read Only Memory (hereinafter referred to as: CD-ROM), a Digital Video Disc Read Only Memory (hereinafter referred to as: DVD-ROM), or other optical media) may be provided. In these cases, each drive may be connected to the bus 18 via one or more data medium interfaces. The memory 28 may include at least one program product having a set (e.g., at least one) of program modules configured to perform the functions of the various embodiments of the present disclosure.

[0155] A program / utility 40 having a set (at least one) of program modules 42 may be stored, for example, in memory 28. Such program modules 42 include, but are not limited to, an operating system, one or more application programs, other program modules, and program data, each of which, or some combination thereof, may include an implementation of a network environment. Program modules 42 generally implement the functions and / or methods of the embodiments described herein.

[0156] The electronic device 6 can also communicate with one or more external devices 14 (e.g., a keyboard, pointing device, display 24, etc.), one or more devices that enable human interaction with the electronic device 6, and / or any device that enables the electronic device 6 to communicate with one or more other computing devices (e.g., a network card, a modem, etc.). This communication can occur via an input / output (I / O) interface 22. Furthermore, the electronic device 6 can communicate with one or more networks (e.g., a local area network (LAN), a wide area network (WAN), and / or a public network such as the Internet) via a network adapter 20. As shown, the network adapter 20 communicates with other modules of the electronic device 6 via a bus 18. It should be understood that, although not shown, other hardware and / or software modules can be used in conjunction with the electronic device 6, including but not limited to microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0157] The processing unit 16 executes various functional applications and parameter information determinations by running programs stored in the memory 28, such as implementing the service accessibility prediction method mentioned in the above embodiments, or implementing the service data acquisition method mentioned in the above embodiments.

[0158] It should be noted that, in the description of this disclosure, the terms "first", "second", etc. are used for descriptive purposes only and should not be understood as indicating or implying relative importance. In addition, in the description of this disclosure, unless otherwise specified, the meaning of "plurality" is two or more.

[0159] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, segment or portion of code that includes one or more executable instructions for implementing the steps of a specific logical function or process, and the scope of the preferred embodiments of the present disclosure includes additional implementations in which functions may be performed out of the order shown or discussed, including performing functions in a substantially simultaneous manner or in the reverse order depending on the functions involved, which should be understood by those skilled in the art to which the embodiments of the present disclosure belong.

[0160] It should be understood that various parts of the present disclosure can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented using software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented using hardware, as in another embodiment, any one of the following technologies known in the art or a combination thereof can be used to implement: a discrete logic circuit having a logic gate circuit for implementing a logic function on a data signal, an application-specific integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0161] Those skilled in the art will understand that all or part of the steps in the method for implementing the above-mentioned embodiment can be completed by instructing related hardware through a program, and the program can be stored in a computer-readable storage medium. When the program is executed, it includes one or a combination of the steps of the method embodiment.

[0162] In addition, the functional units in the various embodiments of the present disclosure may be integrated into a single processing module, or each unit may exist physically separately, or two or more units may be integrated into a single module. The aforementioned integrated modules may be implemented in the form of hardware or in the form of software functional modules. If the integrated modules are implemented in the form of software functional modules and sold or used as independent products, they may also be stored in a computer-readable storage medium.

[0163] The storage medium mentioned above can be a read-only memory, a magnetic disk or an optical disk, etc.

[0164] Throughout this specification, reference to terms such as "one embodiment," "some embodiments," "examples," "specific examples," or "some examples" means that the specific features, structures, materials, or characteristics described in conjunction with that embodiment or example are included in at least one embodiment or example of the present disclosure. In this specification, schematic representations of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.

[0165] Although the embodiments of the present disclosure have been shown and described above, it is understood that the above embodiments are illustrative and are not to be construed as limitations on the present disclosure. A person skilled in the art may change, modify, replace and vary the above embodiments within the scope of the present disclosure.

Claims

1. A method for predicting accessibility, characterized in that: The method comprises: Acquire first service quality data and first behavior data related to the user's access behavior within a first time period; The first quality of service data and the first behavior data are input into a pre-trained accessibility prediction model to obtain an accessibility prediction result of the user in a second time period output by the accessibility prediction model, wherein the accessibility prediction model includes: a target prediction model and a target classification model, the target prediction model is used to predict the second quality of service data and the second behavior data of the user in the second time period, and the target classification model is used to determine the accessibility prediction result based on the second quality of service data and the second behavior data, and the first time period is before the second time period.

2. The method according to claim 1, wherein Inputting the first service quality data and the first behavior data into a pre-trained accessibility prediction model to obtain an accessibility prediction result of the predicted user in the second time period output by the accessibility prediction model includes: Inputting the first service quality data and the first behavior data into the target prediction model to obtain target prediction data output by the target prediction model, wherein the target prediction data includes: second service quality data and second behavior data; The target prediction data is input into the target classification model to obtain the accessibility prediction result output by the target classification model.

3. The method according to claim 2, wherein Inputting the first service quality data and the first behavior data into the target prediction model to obtain target prediction data output by the target prediction model includes: generating a target data sequence according to the first quality of service data and the first behavior data; Dividing the target data sequence based on a preset time step to obtain a plurality of initial data blocks; Inputting each of the initial data blocks into the target prediction model to obtain initial prediction data output by the target prediction model corresponding to each of the initial data blocks; The initial prediction data corresponding to each of the initial data blocks are fused to obtain the target prediction data.

4. The method according to claim 3, wherein Inputting each of the initial data blocks into a target prediction model to obtain initial prediction data output by the target prediction model and corresponding to each of the initial data blocks, respectively, includes: Input each target input data into the i-th processing layer of the target prediction model to obtain i-th layer output data corresponding to each target input data output by the i-th processing layer, wherein i is a positive integer, and when i=1, the target input data is the initial data block, and when i>1, the target input data is the i-th layer output data; The initial prediction data is generated according to the output data of each processing layer in the i processing layers in the target prediction model.

5. The method according to claim 4, wherein Generating the initial prediction data according to the output data of each processing layer in the i processing layers in the target prediction model includes: The output data of each processing layer in the i processing layers in the target prediction model are fused to obtain the initial prediction data.

6. The method according to claim 4, wherein The step of inputting each target input data into the i-th processing layer of the target prediction model to obtain i-th output data output by the i-th processing layer corresponding to each target input data includes: Inputting the target input data into an i-th processing layer of the target prediction model to determine a first query matrix, a first key matrix, and a first value matrix corresponding to the target input data; Determining a first attention weight matrix corresponding to the target input data based on the first query matrix, the first key matrix, and the first value matrix; Generate initial output data according to the first attention weight and the first value matrix; The initial output data is processed based on a nonlinear activation function and the state transition function to obtain the i-th layer output data.

7. The method according to any one of claims 1 to 6, wherein: The target prediction model is obtained by optimizing the initial parameters of the ensemble classification model based on random undersampling based on the differential evolution algorithm to determine the target model parameters, and updating the initial parameters of the ensemble classification model based on random undersampling according to the target model parameters.

8. The method according to any one of claims 1 to 6, wherein: The target prediction model is a long sequence multivariate time series prediction model based on triangular variable specific attention.

9. An accessibility prediction device, characterized in that: The device comprises: An acquisition module, configured to acquire first service quality data and first behavior data related to an access behavior of a user within a first time period; A processing module is configured to input the first quality of service data and the first behavior data into a pre-trained accessibility prediction model to obtain an accessibility prediction result of the predicted user in a second time period output by the accessibility prediction model, wherein the accessibility prediction model includes: a target prediction model and a target classification model, the target prediction model is configured to predict the second quality of service data and the second behavior data of the user in the second time period, and the target classification model is configured to determine the accessibility prediction result based on the second quality of service data and the second behavior data, and the first time period is before the second time period.

10. An electronic device, characterized in that: include: processor; a memory for storing instructions executable by the processor; The processor is configured to execute the instructions to implement the method according to any one of claims 1 to 8.

11. A computer-readable storage medium, when instructions in the computer-readable storage medium are executed by a processor of an electronic device, enables the electronic device to perform the method according to any one of claims 1 to 8.

12. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the method according to any one of claims 1 to 8 is implemented.

Images