Mobile Internet service recommendation method, device, electronic device and storage medium

Abstract

The present invention relates to the field of big data analysis and recommendation technology, and in particular to a method, device, electronic device and storage medium for recommending mobile Internet services. The method of the present invention first obtains the user's Internet service usage record; then, starting from multiple dimensions, multiple dimensional data queues are extracted from the Internet service usage record, and the feature vectors extracted from the multiple dimensional data queues are constructed as a first service feature matrix by means of sliding comparison; then, the first service feature matrix is ​​differentiated, averaged and pooled, and the obtained matrix is ​​sequentially input into an Internet service demand model to obtain multi-dimensional demand; finally, according to the multi-dimensional demand, a target service is selected from multiple mobile Internet services for recommendation. The present invention constructs a feature matrix through service dimension data, and improves the accuracy and effect of recommendation by extracting long-term service demand characteristics and demand change characteristics.

Description

Technical Field

[0001] The present invention relates to the field of big data analysis and recommendation technology, and in particular to a mobile Internet service recommendation method, device, electronic device and storage medium. Background Art

[0002] With the development of the information industry, the Internet, especially the mobile Internet, has been fully integrated into people's lives, and users' demand for Internet resources is increasing. People also hope that mobile operators can provide rich, comprehensive and user-friendly services.

[0003] However, due to the diversity of users, the various packages and traffic packages provided by mobile operators cannot fully meet the needs of different user groups. Some users usually need to purchase traffic packages outside the packages to meet their needs, while avoiding a large amount of extra costs caused by exceeding the traffic outside the packages. Therefore, mobile operators need to consider how to accurately recommend traffic services, while meeting user needs and minimizing user extra expenses, so as to increase user satisfaction with the operator and increase user stickiness on the basis of increasing operator revenue.

[0004] Currently, in terms of customers' Internet needs, major mainstream algorithms make recommendations to users based on popularity and user habits. However, popularity and habits have a certain timeliness, which makes the recommendation results fail to meet user preferences and the recommendation results are not ideal.

[0005] Based on this, it is necessary to develop and design a mobile Internet service recommendation method. Summary of the invention

[0006] The embodiments of the present invention provide a mobile Internet service recommendation method, device, electronic device and storage medium, which are used to solve the problem of unsatisfactory mobile Internet service recommendation in the prior art.

[0007] In a first aspect, an embodiment of the present invention provides a mobile Internet service recommendation method, comprising:

[0008] Obtain users’ Internet service usage records;

[0009] Starting from multiple dimensions, multiple dimension data queues are extracted from the Internet service usage records, and feature vectors extracted from the multiple dimension data queues are constructed into a first service feature matrix by means of sliding comparison, wherein each feature vector corresponds to one dimension data queue;

[0010] Differentiating, averaging and pooling the first service feature matrix, and sequentially inputting the obtained matrix into an Internet service demand model to obtain multi-dimensional demand;

[0011] A target service is selected from a plurality of mobile Internet services for recommendation according to the multi-dimensional requirements.

[0012] In a possible implementation, starting from multiple dimensions, extracting multiple dimensional data queues from Internet service usage records, and constructing the feature vectors extracted from the multiple dimensional data queues into a first service feature matrix by means of sliding comparison, including:

[0013] Acquire multiple dimensional sample sets, where each dimensional sample set corresponds to one dimension, and the dimensional sample set includes multiple sample data queues representing the volatility of dimensional data;

[0014] Sequentially taking out dimension sample sets from the multiple dimension sample sets as the dimension sample sets to be processed;

[0015] A sliding comparison data set is obtained by slidingly comparing the sample data queue of the dimensional sample set to be processed with the target dimensional data queue, wherein the target dimensional data queue is a dimensional data queue corresponding to the dimensional sample set to be processed, and the sliding comparison data set includes a plurality of comparison queues, each comparison queue corresponding to a sample data queue of the dimensional sample set to be processed;

[0016] Performing averaging processing on multiple comparison queues of the slip comparison data set respectively, and constructing the results of the averaging processing as feature vectors;

[0017] According to the dimension corresponding to the dimensional sample set to be processed, the feature vector is added to the first service feature matrix;

[0018] If the traversal of the multiple dimensional sample sets is not completed, the process jumps to the step of sequentially taking out dimensional sample sets from the multiple dimensional sample sets as dimensional sample sets to be processed.

[0019] In a possible implementation, obtaining the sliding comparison data set by slidingly comparing the sample data queue of the to-be-processed dimensional sample set with the target dimensional data queue includes:

[0020] Sequentially taking out sample data queues from the to-be-processed dimensional sample set as the to-be-processed sample data queues;

[0021] Take out data with the same length as the to-be-processed sample data queue from the first position of the target dimension data queue to construct a to-be-compared data queue;

[0022] The comparison values ​​of the data queue to be compared and the sample data queue to be processed are calculated according to a first formula, wherein the first formula is:

[0023]

[0024] In the formula, is the comparison value, The queue of sample data to be processed data, The first data, The total number of data in the sample data queue to be processed;

[0025] Adding the comparison value to the to-be-compared data queue corresponding to the comparison queue;

[0026] If the traversal of the target dimension data queue is not completed, the first position is adjusted according to a predetermined step value, and the process jumps to the step of taking out data having the same length as the queue of sample data to be processed from the first position of the target dimension data queue to construct a queue of data to be compared;

[0027] Adding the to-be-compared data queue corresponding to the comparison queue into the sliding comparison data set;

[0028] If the traversal of the to-be-processed dimensional sample set is not completed, the process jumps to the step of sequentially taking out sample data queues from the to-be-processed dimensional sample set as the to-be-processed sample data queue.

[0029] In a possible implementation manner, performing differentiation, averaging, and pooling on the first service feature matrix includes:

[0030] Acquire multiple operators, wherein the multiple operators include a mean operator and a difference operator;

[0031] Extracting a data block of the same type as the operator from the second position of the first service feature matrix as a data block to be processed;

[0032] The data block to be processed is subjected to operator processing according to the second formula and the multiple operators to obtain multiple operator results, wherein each operator result corresponds to an operator, and the second formula is:

[0033]

[0034] In the formula, is the operatorized result, For operator Line Column data, The data block to be processed Line Column data, is the total number of rows of operators, is the total number of columns of the operator;

[0035] According to the second position, adding the plurality of operatorized results to a plurality of operatorized matrices respectively;

[0036] If the traversal of the first service feature matrix is ​​not completed, the second position is offset according to a predetermined step length, and the process jumps to the step of extracting a data block of the same type as the operator from the second position of the first service feature matrix as a data block to be processed;

[0037] Performing maximum pooling on the multiple operator matrices respectively to obtain multiple first pooling matrices;

[0038] Normalization is performed on the multiple first pooling matrices respectively to obtain multiple second service feature matrices.

[0039] In a possible implementation, the Internet service demand model is constructed based on multiple sample service feature sets, wherein each sample service feature set corresponds to one user, and the service feature set includes multiple sample second service feature matrices and multiple sample dimensional requirements. The construction process of the Internet service demand model includes:

[0040] Acquire a demand basic model and a plurality of coefficient arrays, wherein the demand basic model has a plurality of undetermined parameters, and the number of data in the coefficient array is the same as the number of the plurality of undetermined parameters;

[0041] Substituting the multiple coefficient arrays into the basic model respectively to obtain multiple test models;

[0042] traversally extracting a feature set from the plurality of sample service feature sets as a feature set to be processed;

[0043] Inputting the plurality of sample second service feature matrices of the feature set to be processed into the plurality of test models respectively to obtain a plurality of dimension demand data sets, wherein each dimension demand data set corresponds to a test model, and the dimension demand data set includes a plurality of dimension demand indication values;

[0044] Determine the fitness of each test model according to each dimension requirement data set and the sample dimension requirement of the feature set to be processed, thereby obtaining multiple fitnesses;

[0045] Selecting three best models from the multiple test models according to the multiple fitness values;

[0046] If there is no applicable model among the three best models, then adjust other models among the multiple test models according to the three best models, and jump to the step of erroneously extracting feature sets from the multiple sample service feature sets as feature sets to be processed, wherein the applicable model is a test model whose fitness is less than a fitness threshold in a continuous preset number of iterations;

[0047] Otherwise, the applicable model is taken as the Internet service demand model.

[0048] In a possible implementation, the demand base model is:

[0049]

[0050] In the formula, For the Dimensional requirements, For the The first parameter, is the variable, is the total number of exponentials, For the Bias parameters, For the The second parameter, Substitute the matrix Line The elements of the column, is the total number of rows in the matrix, is the total number of columns of the matrix, For the The third parameter, is the total number of input matrices, For the The input matrix Line Elements of a column.

[0051] In a possible implementation manner, adjusting other models in the multiple test models according to the three best models includes:

[0052] According to the third formula and the three best models, other models in the multiple test models are adjusted, wherein the third formula is:

[0053]

[0054] In the formula, For the The adjusted test model parameters, For the The adjusted test model parameters, , , as well as They are the first distance, the second distance, the third distance and the fourth distance respectively. , , as well as They are the fourth parameter, the fifth parameter, the sixth parameter and the seventh parameter, , , as well as They are the optimal test models. parameters, suboptimal test model parameters, the third best test model The optimal parameters of the current test model in all previous iterations. parameters.

[0055] In a second aspect, an embodiment of the present invention provides a mobile Internet service recommendation device, which is used to implement the mobile Internet service recommendation method described in the first aspect or any possible implementation of the first aspect, wherein the mobile Internet service recommendation device includes:

[0056] A record acquisition module is used to acquire the user's Internet service usage records;

[0057] A feature matrix construction module is used to extract multiple dimensional data queues from the Internet service usage records from multiple dimensions, and construct the feature vectors extracted from the multiple dimensional data queues into a first service feature matrix by means of sliding comparison, wherein each feature vector corresponds to one dimensional data queue;

[0058] A dimensional demand determination module, used for performing differentiation, averaging and pooling on the first service feature matrix, and sequentially inputting the obtained matrix into an Internet service demand model to obtain multi-dimensional demand;

[0059] as well as,

[0060] The mobile Internet service recommendation module is used to select a target service from a plurality of mobile Internet services for recommendation according to the multi-dimensional demand.

[0061] In a third aspect, an embodiment of the present invention provides an electronic device, comprising a memory and a processor, wherein the memory stores a computer program that can be run on the processor, and when the processor executes the computer program, it implements the steps of the method described in the first aspect or any possible implementation method of the first aspect.

[0062] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the method described in the first aspect or any possible implementation of the first aspect are implemented.

[0063] Compared with the prior art, the embodiments of the present invention have the following beneficial effects:

[0064] The embodiment of the present invention discloses a method for recommending mobile Internet services. The embodiment of the method for recommending mobile Internet services of the present invention first obtains the Internet service usage record of the user; then, starting from multiple dimensions, multiple dimensional data queues are extracted from the Internet service usage record, and the feature vectors extracted from the multiple dimensional data queues are constructed as a first service feature matrix by means of sliding comparison, wherein each feature vector corresponds to a dimensional data queue; then, the first service feature matrix is ​​differentiated, averaged, and pooled, and the obtained matrix is ​​sequentially input into the Internet service demand model to obtain multi-dimensional demand; finally, according to the multi-dimensional demand, a target service is selected from multiple mobile Internet services for recommendation. The embodiment of the present invention constructs a feature matrix representing multiple dimensions of services through service dimension data, and extracts long-term service demand characteristics and demand change characteristics by performing mean and difference processing on the matrix. In other words, the long-term demand characteristics and medium- and short-term demand volatility are comprehensively considered, and service dimension demand recommendations are made based on the above characteristics, thereby improving the accuracy and effect of the recommendation. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0066] Figure 1 is a flow chart of a mobile Internet service recommendation method provided by an embodiment of the present invention;

[0067] Figure 2 is a schematic diagram of a feature vector construction process provided by an embodiment of the present invention;

[0068] Figure 3 It is a principle diagram of the operator processing process of the first service feature matrix provided by an embodiment of the present invention;

[0069] Figure 4 It is a schematic diagram of the process of constructing an Internet service demand model provided by an embodiment of the present invention;

[0070] Figure 5 is a functional block diagram of a mobile Internet service recommendation device provided by an embodiment of the present invention;

[0071] Figure 6 It is a functional block diagram of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0072] In the following description, specific details such as specific system structures and technologies are provided for the purpose of illustration rather than limitation so as to provide a thorough understanding of the embodiments of the present invention. However, it should be clear to those skilled in the art that the present invention may be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, and methods are omitted so as not to obscure the description of the present invention with unnecessary details.

[0073] In order to make the purpose, technical solutions and advantages of the present invention more clear, a specific implementation method will be described below in conjunction with the accompanying drawings.

[0074] The following is a detailed description of an embodiment of the present invention. This example is implemented based on the technical solution of the present invention, and provides a detailed implementation method and a specific operation process, but the protection scope of the present invention is not limited to the following embodiments.

[0075] Figure 1 A flowchart of a mobile Internet service recommendation method provided in an embodiment of the present invention.

[0076] like Figure 1 As shown, it shows a flowchart of the implementation method of the mobile Internet service recommendation method provided by the embodiment of the present invention, which is described in detail as follows:

[0077] In step 101, the user's Internet service usage record is obtained.

[0078] In step 102, starting from multiple dimensions, multiple dimensional data queues are extracted from the Internet service usage records, and the feature vectors extracted from the multiple dimensional data queues are constructed into a first service feature matrix by sliding comparison, wherein each feature vector corresponds to a dimensional data queue.

[0079] In some implementations, step 102 includes:

[0080] Acquire multiple dimensional sample sets, where each dimensional sample set corresponds to one dimension, and the dimensional sample set includes multiple sample data queues representing the volatility of dimensional data;

[0081] Sequentially taking out dimension sample sets from the multiple dimension sample sets as the dimension sample sets to be processed;

[0082] A sliding comparison data set is obtained by slidingly comparing the sample data queue of the dimensional sample set to be processed with the target dimensional data queue, wherein the target dimensional data queue is a dimensional data queue corresponding to the dimensional sample set to be processed, and the sliding comparison data set includes a plurality of comparison queues, each comparison queue corresponding to a sample data queue of the dimensional sample set to be processed;

[0083] Performing averaging processing on multiple comparison queues of the slip comparison data set respectively, and constructing the results of the averaging processing as feature vectors;

[0084] According to the dimension corresponding to the dimensional sample set to be processed, the feature vector is added to the first service feature matrix;

[0085] If the traversal of the multiple dimensional sample sets is not completed, the process jumps to the step of sequentially taking out dimensional sample sets from the multiple dimensional sample sets as dimensional sample sets to be processed.

[0086] In some implementations, obtaining the sliding comparison data set by slidingly comparing the sample data queue of the to-be-processed dimensional sample set with the target dimensional data queue includes:

[0087] Sequentially taking out sample data queues from the to-be-processed dimensional sample set as the to-be-processed sample data queues;

[0088] Take out data with the same length as the to-be-processed sample data queue from the first position of the target dimension data queue to construct a to-be-compared data queue;

[0089] The comparison values ​​of the data queue to be compared and the sample data queue to be processed are calculated according to a first formula, wherein the first formula is:

[0090]

[0091] In the formula, is the comparison value, The queue of sample data to be processed data, The first data, The total number of data in the sample data queue to be processed;

[0092] Adding the comparison value to the to-be-compared data queue corresponding to the comparison queue;

[0093] If the traversal of the target dimension data queue is not completed, the first position is adjusted according to a predetermined step value, and the process jumps to the step of taking out data having the same length as the queue of sample data to be processed from the first position of the target dimension data queue to construct a queue of data to be compared;

[0094] Adding the to-be-compared data queue corresponding to the comparison queue into the sliding comparison data set;

[0095] If the traversal of the to-be-processed dimensional sample set is not completed, the process jumps to the step of sequentially taking out sample data queues from the to-be-processed dimensional sample set as the to-be-processed sample data queue.

[0096] Exemplarily, the embodiments of the present invention mine the long-term and recent service conditions of users through Internet service usage records, and the starting point is to take into account the long-term service selection characteristics and the volatility of recent service selection. On the other hand, the characteristics of services have multiple dimensions, for example, service dimensions include: Internet usage time, Internet service applications, Internet data categories (calls, audio, video or images), Internet usage locations, etc. These dimensions are the focus of users and the focus that affects user choices. Each service corresponds to the above multiple dimensions.

[0097] By aggregating multiple services of the user according to dimensions, a dimensional data queue corresponding to the dimension is obtained. Since the lengths of the data queues are different and the data are relatively scattered and contain multiple features, the implementation method of the present invention extracts features related to service selection from these data queues and sends them into the mobile Internet service recommendation model to obtain recommended service dimensions.

[0098] like Figure 2 As shown, in order to achieve the above purpose, after constructing the dimension data queue, the embodiment of the present invention extracts the features of the dimension data queue through multiple dimension sample sets 201, thereby obtaining the feature vector of the dimension. The dimension sample set 201 includes multiple sample data queues 202, and each sample data queue 202 is respectively constructed by sliding with the dimension data queue 203 to construct a sliding comparison data set 205. After obtaining the complete sliding comparison data set 205, the sliding comparison data set 205 can be averaged to obtain the averaged result 206. The averaged result 206 is added to the feature vector corresponding to the dimension sample set 201 (the sample data queue 202 belongs to the dimension sample set 201) because it corresponds to the sample data queue 202.

[0099] One sliding process is to sequentially shift and extract the to-be-compared data queue 204 of the same length as the sample data queue 202 from the dimension data queue 203, and calculate the comparison value between the two by the first formula:

[0100]

[0101] In the formula, is the comparison value, The queue of sample data to be processed data, The first data, The total number of data in the sample data queue to be processed.

[0102] It can be seen that each translation will obtain a comparison value, and multiple comparison values ​​together constitute the sliding comparison data set 205.

[0103] Here we need to explain a way to construct a sample data queue. A recommended construction method is to extract multiple historical dimension data queues, intercept data queues from them for clustering, and construct the data queue at the center of the cluster as the initial queue. These initial queues are unitized to obtain the sample data queue.

[0104] In addition, another way is to construct multiple orthogonal waveforms through the fundamental waves of wavelet transform, and discretize these waveforms to obtain sample data queues. Compared with the sample data queues obtained by clustering, the sample data queues obtained by clustering are obviously more suitable for feature extraction scenarios.

[0105] In summary, it can be clearly known through the above steps that each dimensional data queue obtains a feature vector through the corresponding dimensional sample set, and multiple dimensional data queues of the service are composed of multiple feature vectors to obtain a first service feature matrix.

[0106] In step 103, the first service feature matrix is ​​differentiated, averaged and pooled, and the obtained matrix is ​​sequentially input into the Internet service demand model to obtain multi-dimensional demand.

[0107] In some implementations, performing differentiation, averaging, and pooling on the first service feature matrix includes:

[0108] Acquire multiple operators, wherein the multiple operators include a mean operator and a difference operator;

[0109] Extracting a data block of the same type as the operator from the second position of the first service feature matrix as a data block to be processed;

[0110] The data block to be processed is subjected to operator processing according to the second formula and the multiple operators to obtain multiple operator results, wherein each operator result corresponds to an operator, and the second formula is:

[0111]

[0112] In the formula, is the operatorized result, For operator Line Column data, The data block to be processed Line Column data, is the total number of rows of operators, is the total number of columns of the operator;

[0113] According to the second position, adding the plurality of operatorized results to a plurality of operatorized matrices respectively;

[0114] If the traversal of the first service feature matrix is ​​not completed, the second position is offset according to a predetermined step length, and the process jumps to the step of extracting a data block of the same type as the operator from the second position of the first service feature matrix as a data block to be processed;

[0115] Performing maximum pooling on the multiple operator matrices respectively to obtain multiple first pooling matrices;

[0116] Normalization is performed on the multiple first pooling matrices respectively to obtain multiple second service feature matrices.

[0117] In some embodiments, the Internet service demand model is constructed based on multiple sample service feature sets, wherein each sample service feature set corresponds to one user, and the service feature set includes multiple sample second service feature matrices and multiple sample dimensional requirements. The construction process of the Internet service demand model includes:

[0118] Acquire a demand basic model and a plurality of coefficient arrays, wherein the demand basic model has a plurality of undetermined parameters, and the number of data in the coefficient array is the same as the number of the plurality of undetermined parameters;

[0119] Substituting the multiple coefficient arrays into the basic model respectively to obtain multiple test models;

[0120] traversally extracting a feature set from the plurality of sample service feature sets as a feature set to be processed;

[0121] Inputting the plurality of sample second service feature matrices of the feature set to be processed into the plurality of test models respectively to obtain a plurality of dimension demand data sets, wherein each dimension demand data set corresponds to a test model, and the dimension demand data set includes a plurality of dimension demand indication values;

[0122] Determine the fitness of each test model according to each dimension requirement data set and the sample dimension requirement of the feature set to be processed, thereby obtaining multiple fitnesses;

[0123] Selecting three best models from the multiple test models according to the multiple fitness values;

[0124] If there is no applicable model among the three best models, then adjust other models among the multiple test models according to the three best models, and jump to the step of erroneously extracting feature sets from the multiple sample service feature sets as feature sets to be processed, wherein the applicable model is a test model whose fitness is less than a fitness threshold in a continuous preset number of iterations;

[0125] Otherwise, the applicable model is taken as the Internet service demand model.

[0126] In some embodiments, the demand-based model is:

[0127]

[0128] In the formula, For the Dimensional requirements, For the The first parameter, is the variable, is the total number of exponentials, For the Bias parameters, For the The second parameter, Substitute the matrix Line The elements of the column, is the total number of rows in the matrix, is the total number of columns of the matrix, For the The third parameter, is the total number of input matrices, For the The input matrix Line Elements of a column.

[0129] In some embodiments, adjusting other models in the plurality of test models according to the three best models comprises:

[0130] According to the third formula and the three best models, other models in the multiple test models are adjusted, wherein the third formula is:

[0131]

[0132] In the formula, For the The adjusted test model parameters, For the The adjusted test model parameters, , , as well as They are the first distance, the second distance, the third distance and the fourth distance respectively. , , as well as They are the fourth parameter, the fifth parameter, the sixth parameter and the seventh parameter, , , as well as They are the optimal test models. parameters, suboptimal test model parameters, the third best test model The optimal parameters of the current test model in all previous iterations. parameters.

[0133] Exemplarily, the features contained in the matrix obtained through the aforementioned steps still contain some features that have little impact on the recommendation results. For example, the features with very small values ​​obtained through the aforementioned steps. In order to further compress the features, reduce the dimension of the data, and improve the representativeness of the features, after obtaining the first service feature matrix, the implementation mode of the present invention further performs averaging, differentiation and pooling processing on the first service feature matrix.

[0134] The averaging and differentiation are respectively performed on the first service feature matrix by corresponding operators. For example, a 3×3 averaging operator is:

[0135]

[0136] A 3×3 difference operator is:

[0137]

[0138] The matrix obtained by processing the above two operators is then pooled to obtain a matrix, and is input into the Internet service demand model in sequence to obtain the dimensional demand data of the recommended service (for example, the transportation distance, transportation method, transportation time, etc. of the recommended service).

[0139] like Figure 3 As shown, the figure shows a flow diagram of operator processing. Data blocks 303 of the same dimension as the operator 301 are taken out from the first service feature matrix 302 in turn, and the two are operated. The operation results are stored in the operator matrix 304 according to the position of the data block 303 in the first service feature matrix 302. After traversing the operator matrix 302, the operator matrix 304 of the corresponding operator is obtained. Each operator matrix is ​​pooled separately, for example, the maximum pooling obtains the first pooling matrix. The first pooling matrix is ​​then normalized to obtain the second service feature matrix.

[0140] One operator-based calculation method is to use the second formula:

[0141]

[0142] In the formula, is the operatorized result, For operator Line Column data, The data block to be processed Line Column data, is the total number of rows of operators, is the total number of columns of the operator.

[0143] The obtained multiple second service feature matrices are input into the model in a predetermined order to obtain the dimensional requirements of the recommended service.

[0144] The Internet service demand model is actually based on the basic demand model. It is constructed by repeatedly iterating multiple parameters in the basic demand model using multiple sample service feature sets.

[0145] The demand-based model used in the implementation of the present invention is:

[0146]

[0147] In the formula, For the Dimensional requirements, For the The first parameter, is the variable, is the total number of exponentials, For the Bias parameters, For the The second parameter, Substitute the matrix Line The elements of the column, is the total number of rows in the matrix, is the total number of columns of the matrix, For the The third parameter, is the total number of input matrices, For the The input matrix Line Elements of a column.

[0148] The sample service feature set is a data set constructed based on the characteristics of historical users, including multiple sample second service feature matrices and multiple sample dimension requirements. The construction process of the sample second service feature matrix is ​​the same as that of the second service feature matrix, and each sample dimension requirement corresponds to a sample second service feature matrix, which can be regarded as the actual sharing demand based on the corresponding previous data.

[0149] Figure 4The process of determining model parameters according to the sample service feature set is shown. First, multiple coefficient arrays 401 are obtained and initialized. These coefficient arrays 401 are respectively substituted into the demand basic model 402 to obtain multiple test models 403. At this time, the sample second service feature matrix 404 in the sample service feature set is respectively substituted into the multiple test models 403, and the obtained multiple results 406 (the results are obtained by the output of the test model 403) are compared with the sample dimension requirements 405 in the sample service feature set. At this time, it is found that some test models 403 have small deviations. The best three are used as references to modify the parameters of the other multiple test models 403. One modification method uses the third formula, which is:

[0150]

[0151] In the formula, For the The adjusted test model parameters, For the The adjusted test model parameters, , , as well as They are the first distance, the second distance, the third distance and the fourth distance respectively. , , as well as They are the fourth parameter, the fifth parameter, the sixth parameter and the seventh parameter, , , as well as They are the optimal test models. parameters, suboptimal test model parameters, the third best test model The optimal parameters of the current test model in all previous iterations. parameters.

[0152] After several rounds of iterations, there are one or more test models 403 that have undergone multiple rounds of iterations, whose parameters have not been adjusted and whose deviations are within expectations. At this time, the best test model 403 is used as the Internet service demand model.

[0153] In step 104, a target service is selected from a plurality of mobile Internet services for recommendation according to the multi-dimensional demand.

[0154] Exemplarily, the dimension requirements obtained according to the above process are searched and compared from the database, and the service whose service dimension data is closest to the dimension requirement is selected and recommended to the user.

[0155] In one application scenario, the mobile Internet service data of users in a scenic spot is analyzed. After the dimensional feature analysis, it is shown that some users in the scenic spot use mobile Internet services during their daily activities and use audio, video and calls. Based on this demand, it can be recommended that these users use voice call services. After the dimensional feature analysis, it is shown that another part of users use mobile Internet services during the evening hours and use entertainment traffic data. The Internet service application is entertainment application. Based on this demand, it can be recommended that they use directional data traffic services.

[0156] In another application scenario, we analyze users and extract multi-dimensional features through their mobile Internet service data usage records. The main dimensional features are concentrated on the fact that the data activity period is the holiday period, the mobile Internet data usage locations are mostly scenic spots, the mobile Internet data types are mostly audio and video, and calls, and the mobile Internet applications used are navigation and consumption applications. It can be seen that the user's mobile Internet service data demand is travel data services, so we can recommend holiday data services for them.

[0157] The implementation method of the mobile Internet service recommendation method of the present invention first obtains the user's Internet service usage record; then, starting from multiple dimensions, extracts multiple dimensional data queues from the Internet service usage record, and constructs the feature vectors extracted from the multiple dimensional data queues into a first service feature matrix by means of sliding comparison, wherein each feature vector corresponds to a dimensional data queue; then, the first service feature matrix is ​​differentiated, averaged, and pooled, and the obtained matrix is ​​sequentially input into the Internet service demand model to obtain multi-dimensional demand; finally, according to the multi-dimensional demand, the target service is selected from multiple mobile Internet services for recommendation. The implementation method of the present invention constructs a feature matrix representing multiple dimensions of the service through service dimension data, and extracts long-term service demand characteristics and demand change characteristics by performing mean and difference processing on the matrix. In other words, it comprehensively considers long-term demand characteristics and medium- and short-term demand volatility, and recommends service dimension demand based on the above characteristics, thereby improving the accuracy and effect of the recommendation.

[0158] The embodiment of the present invention undergoes a first round of feature extraction and a second round of difference and mean feature extraction, and the resulting matrix has a small dimension, reduces the complexity of the model, and reduces the difficulty of model application. In addition, through the above process, the main features that affect service demand are retained and refined, thereby ensuring the accuracy of recommendations.

[0159] It should be understood that the size of the serial numbers of the steps in the above implementation does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the implementation method of the present invention.

[0160] The following is an implementation of the device of the present invention. For details not described in detail, reference may be made to the corresponding method implementation described above.

[0161] Figure 5 is a functional block diagram of a mobile Internet service recommendation device provided by an embodiment of the present invention, referring to Figure 5 The mobile Internet service recommendation device includes: a record acquisition module 501, a feature matrix construction module 502, a dimension requirement determination module 503 and a mobile Internet service recommendation module 504, wherein:

[0162] Record acquisition module 501, used to acquire the user's Internet service usage record;

[0163] The feature matrix construction module 502 is used to extract multiple dimension data queues from the Internet service usage records from multiple dimensions, and construct the feature vectors extracted from the multiple dimension data queues into a first service feature matrix by means of sliding comparison, wherein each feature vector corresponds to one dimension data queue;

[0164] A dimensional demand determination module 503 is used to perform differentiation, averaging and pooling on the first service feature matrix, and sequentially input the obtained matrix into an Internet service demand model to obtain multi-dimensional demand;

[0165] as well as,

[0166] The mobile Internet service recommendation module 504 is used to select a target service from a plurality of mobile Internet services for recommendation according to the multi-dimensional demand.

[0167] Figure 6 is a functional block diagram of an electronic device provided by an embodiment of the present invention. Figure 6 As shown, the electronic device 6 of this embodiment includes: a processor 600 and a memory 601, wherein the memory 601 stores a computer program 602 that can be run on the processor 600. When the processor 600 executes the computer program 602, the steps in the above-mentioned mobile Internet service recommendation method and the embodiment are implemented, for example Figure 1 Steps 101 to 104 are shown.

[0168] Exemplarily, the computer program 602 may be divided into one or more modules / units, and the one or more modules / units are stored in the memory 601 and executed by the processor 600 to implement the present invention.

[0169] The electronic device 6 may be a computing device such as a desktop computer, a notebook, a PDA, or a cloud server. The electronic device 6 may include, but is not limited to, a processor 600 and a memory 601. Those skilled in the art will appreciate that Figure 6 It is only an example of the electronic device 6 and does not constitute a limitation of the electronic device 6. It may include more or fewer components than shown in the figure, or a combination of certain components, or different components. For example, the electronic device 6 may also include input and output devices, network access devices, buses, etc.

[0170] The processor 600 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc.

[0171] The memory 601 may be an internal storage unit of the electronic device 6, such as a hard disk or memory of the electronic device 6. The memory 601 may also be an external storage device of the electronic device 6, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the electronic device 6. Further, the memory 601 may also include both an internal storage unit and an external storage device of the electronic device 6. The memory 601 is used to store the computer program 602 and other programs and data required by the electronic device 6. The memory 601 may also be used to temporarily store data that has been output or is to be output.

[0172] Those skilled in the art can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In actual applications, the above-mentioned function allocation can be completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the implementation method can be integrated into a processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of this application. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method implementation method, which will not be repeated here.

[0173] In the above-mentioned embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0174] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0175] In the embodiments provided by the present invention, it should be understood that the disclosed devices / electronic devices and methods can be implemented in other ways. For example, the device / electronic device embodiments described above are only schematic. For example, the division of the modules or units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0176] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0177] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0178] If the integrated module / unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present invention implements all or part of the processes in the above-mentioned implementation method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by the processor, the steps of the above-mentioned various methods and device implementation methods can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may include: any entity or device that can carry the computer program code, recording medium, U disk, mobile hard disk, disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal and software distribution medium.

[0179] The above-described embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, a person skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention, and should all be included in the protection scope of the present invention.

Claims

1. A mobile Internet service recommendation method, characterized in that: include: Obtain users’ Internet service usage records; Starting from multiple dimensions, extracting multiple dimensional data queues from the Internet service usage records, and constructing a first service feature matrix by sliding comparison using feature vectors extracted from the multiple dimensional data queues, wherein each feature vector corresponds to one dimensional data queue; Differentiating, averaging and pooling the first service feature matrix, and sequentially inputting the obtained matrix into an Internet service demand model to obtain multi-dimensional demand; Selecting a target service from a plurality of mobile Internet services for recommendation according to the multi-dimensional requirements; in, The method starts from multiple dimensions, extracts multiple dimensional data queues from the Internet service usage records, and constructs the feature vectors extracted from the multiple dimensional data queues into a first service feature matrix by means of sliding comparison, including: Acquire multiple dimensional sample sets, where each dimensional sample set corresponds to one dimension, and the dimensional sample set includes multiple sample data queues representing the volatility of dimensional data; Sequentially taking out dimension sample sets from the multiple dimension sample sets as the dimension sample sets to be processed; A sliding comparison data set is obtained by slidingly comparing the sample data queue of the dimensional sample set to be processed with the target dimensional data queue, wherein the target dimensional data queue is a dimensional data queue corresponding to the dimensional sample set to be processed, and the sliding comparison data set includes a plurality of comparison queues, each comparison queue corresponding to a sample data queue of the dimensional sample set to be processed; Performing averaging processing on multiple comparison queues of the slip comparison data set respectively, and constructing the results of the averaging processing as feature vectors; According to the dimension corresponding to the dimensional sample set to be processed, the feature vector is added to the first service feature matrix; If the traversal of the multiple dimensional sample sets is not completed, jump to the step of sequentially taking out dimensional sample sets from the multiple dimensional sample sets as dimensional sample sets to be processed; in, The Internet service demand model is constructed based on multiple sample service feature sets, wherein each sample service feature set corresponds to one user, and the service feature set includes multiple sample second service feature matrices and multiple sample dimensional requirements. The construction process of the Internet service demand model includes: Acquire a demand basic model and a plurality of coefficient arrays, wherein the demand basic model has a plurality of undetermined parameters, and the number of data in the coefficient array is the same as the number of the plurality of undetermined parameters; Substituting the multiple coefficient arrays into the basic model respectively to obtain multiple test models; traversally extracting a feature set from the plurality of sample service feature sets as a feature set to be processed; Inputting the plurality of sample second service feature matrices of the feature set to be processed into the plurality of test models respectively to obtain a plurality of dimension demand data sets, wherein each dimension demand data set corresponds to a test model, and the dimension demand data set includes a plurality of dimension demand indication values; Determine the fitness of each test model according to each dimension requirement data set and the sample dimension requirement of the feature set to be processed, thereby obtaining multiple fitnesses; Selecting three best models from the multiple test models according to the multiple fitness values; If there is no applicable model among the three best models, then adjust other models among the multiple test models according to the three best models, and jump to the step of erroneously extracting feature sets from the multiple sample service feature sets as feature sets to be processed, wherein the applicable model is a test model whose fitness is less than a fitness threshold in a continuous preset number of iterations; Otherwise, the applicable model is taken as the Internet service demand model.

2. The mobile Internet service recommendation method according to claim 1, characterized in that: The method of obtaining a sliding comparison data set by slidingly comparing the sample data queue of the to-be-processed dimensional sample set with the target dimensional data queue includes: Sequentially taking out sample data queues from the to-be-processed dimensional sample set as the to-be-processed sample data queues; Take out data with the same length as the to-be-processed sample data queue from the first position of the target dimension data queue to construct a to-be-compared data queue; The comparison values ​​of the data queue to be compared and the sample data queue to be processed are calculated according to a first formula, wherein the first formula is: In the formula, is the comparison value, The queue of sample data to be processed data, The first data, The total number of data in the sample data queue to be processed; Adding the comparison value to the to-be-compared data queue corresponding to the comparison queue; If the traversal of the target dimension data queue is not completed, the first position is adjusted according to a predetermined step value, and the process jumps to the step of taking out data having the same length as the queue of sample data to be processed from the first position of the target dimension data queue to construct a queue of data to be compared; Adding the to-be-compared data queue corresponding to the comparison queue into the sliding comparison data set; If the traversal of the to-be-processed dimensional sample set is not completed, the process jumps to the step of sequentially taking out sample data queues from the to-be-processed dimensional sample set as the to-be-processed sample data queue.

3. The mobile Internet service recommendation method according to claim 1, characterized in that: The performing differentiation, averaging and pooling on the first service feature matrix includes: Acquire multiple operators, wherein the multiple operators include a mean operator and a difference operator; Extracting a data block of the same type as the operator from the second position of the first service feature matrix as a data block to be processed; The data block to be processed is subjected to operator processing according to the second formula and the multiple operators to obtain multiple operator results, wherein each operator result corresponds to an operator, and the second formula is: In the formula, is the operatorized result, For operator Line Column data, The data block to be processed Line Column data, is the total number of rows of operators, is the total number of columns of the operator; According to the second position, adding the plurality of operatorized results to a plurality of operatorized matrices respectively; If the traversal of the first service feature matrix is ​​not completed, the second position is offset according to a predetermined step length, and the process jumps to the step of extracting a data block of the same type as the operator from the second position of the first service feature matrix as a data block to be processed; Performing maximum pooling on the multiple operator matrices respectively to obtain multiple first pooling matrices; Normalization is performed on the multiple first pooling matrices respectively to obtain multiple second service feature matrices.

4. The mobile Internet service recommendation method according to claim 1, characterized in that: The demand-based model is: In the formula, For the Dimensional requirements, For the The first parameter, is the variable, is the total number of exponentials, For the Bias parameters, For the The second parameter, Substitute the matrix Line The elements of the column, is the total number of rows in the matrix, is the total number of columns of the matrix, For the The third parameter, is the total number of input matrices, For the The input matrix Line Elements of a column.

5. The mobile Internet service recommendation method according to claim 1, characterized in that: The adjusting other models in the plurality of test models according to the three best models comprises: According to the third formula and the three best models, other models in the multiple test models are adjusted, wherein the third formula is: In the formula, For the The adjusted test model parameters, For the The adjusted test model parameters, , , as well as They are the first distance, the second distance, the third distance and the fourth distance respectively. , , as well as They are the fourth parameter, the fifth parameter, the sixth parameter and the seventh parameter, , , as well as They are the optimal test models. parameters, suboptimal test model parameters, the third best test model The optimal parameters of the current test model in all previous iterations. parameters.

6. A mobile Internet service recommendation device, characterized in that: Used to implement the mobile Internet service recommendation method according to any one of claims 1 to 5, the mobile Internet service recommendation device comprises: A record acquisition module is used to obtain the user's Internet service usage record; A feature matrix construction module is used to extract multiple dimensional data queues from the Internet service usage records from multiple dimensions, and construct the feature vectors extracted from the multiple dimensional data queues into a first service feature matrix by means of sliding comparison, wherein each feature vector corresponds to one dimensional data queue; A dimensional demand determination module, used for performing differentiation, averaging and pooling on the first service feature matrix, and sequentially inputting the obtained matrix into an Internet service demand model to obtain multi-dimensional demand; as well as, The mobile Internet service recommendation module is used to select a target service from a plurality of mobile Internet services for recommendation according to the multi-dimensional demand.

7. An electronic device comprising a memory and a processor, wherein the memory stores a computer program that can be run on the processor, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 5 are implemented.

8. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.

Images