Information sending channel determination method, apparatus and device, and computer storage medium

Abstract

The invention discloses an information sending channel determination method and device, equipment and a computer storage medium, and is applied to the field of big data. The method comprises the following steps: acquiring user data information, and inputting the user data information into a target model to obtain a feature matrix corresponding to the user data information; and carrying out dimension reduction processing on the feature matrix to obtain a one-dimensional vector. And performing linear combination on the second matrix, a preset offset vector and the one-dimensional vector to obtain a target feature vector. And determining multiple pieces of probability information sent through an information sending channel according to the target feature vector, selecting the maximum value in the multiple pieces of probability information as target probability information, and sending satisfaction survey information to the user through the information sending channel corresponding to the target probability information. By extracting the feature matrix of the user data and performing dimensionality reduction and transformation based on the feature matrix, the information of the information sending channel corresponds to the features of the user data information, so that the matching degree of the information sending channel and the user is improved.

Description

Technical Field

[0001] The present invention relates to the field of big data, and in particular to a method, device, equipment, computer storage medium and computer program product for determining an information sending channel. Background Art

[0002] Currently, the service industry uses various means to improve service quality in order to gain the favor of users. For example, after users complete their business, the company sends satisfaction survey information to users through the backend server to improve service quality based on user feedback.

[0003] However, in the prior art, satisfaction survey information is often sent to users in a single survey method such as SMS questionnaires, WeChat questionnaires, etc., and users may be disgusted by the survey method, resulting in a low match between the information sending channel and the user. Summary of the invention

[0004] The embodiments of the present disclosure provide a method, apparatus, device, computer storage medium, and computer program product for determining an information sending channel, which can improve the matching degree between the information sending channel and the user.

[0005] In a first aspect, an embodiment of the present disclosure provides a method for determining an information sending channel, the method comprising:

[0006] Obtain user data information, including satisfaction evaluation channel information, satisfaction evaluation status information, satisfaction survey authorization information and complaint information;

[0007] Inputting the user data information into the target model, and using the target model to multiply the user data information by the first matrix to obtain a feature matrix corresponding to the user data information;

[0008] Reduce the dimension of the feature matrix to obtain a one-dimensional vector;

[0009] A target feature vector is obtained by linearly combining the second matrix and the preset bias vector with the one-dimensional vector, wherein the second matrix corresponds to the information sending channel;

[0010] Based on the correspondence between the feature vector and the probability information, multiple probabilities of the information being sent through the sending channel are determined according to the target feature vector;

[0011] Select the maximum value among multiple probability information as the target probability information;

[0012] The satisfaction survey information is sent to the user through the information sending channel corresponding to the target probability information.

[0013] In one practicable implementation, obtaining user data information includes:

[0014] Get initial user data information;

[0015] Deleting missing values, duplicate values, and abnormal values ​​of the initial user data information to obtain first user data information;

[0016] Converting non-numeric data in the first user data information into numeric data to obtain second user data information;

[0017] The second user data information is normalized to obtain user data information.

[0018] In an achievable implementation, the user data information is multiplied by the first matrix to obtain a feature matrix corresponding to the user data information, including:

[0019] Constructing a first zero matrix, wherein the number of rows of the first zero matrix is ​​equal to the number of rows of the first matrix, and the number of columns of the first zero matrix is ​​equal to the number of columns of the first matrix;

[0020] Dividing the user data information into a plurality of user data submodules;

[0021] Multiplying the first matrix by the plurality of user data submodules respectively according to a first preset order to obtain a plurality of sub-data;

[0022] A plurality of sub-data are input into a first zero matrix according to a first preset order to obtain a characteristic matrix.

[0023] In one feasible implementation, the feature matrix is ​​reduced in dimension to obtain a one-dimensional vector, including:

[0024] Constructing a second zero matrix, wherein the number of rows of the second zero matrix is ​​less than the number of rows of the characteristic matrix, and the number of columns of the second zero matrix is ​​less than the number of columns of the characteristic matrix;

[0025] Divide the feature matrix into multiple feature submodules;

[0026] Selecting maximum values ​​of multiple feature submodules respectively according to a second preset order to obtain multiple feature maximum values;

[0027] Inputting the plurality of feature maximum values ​​into the second zero matrix according to a second preset order to obtain an initial dimension reduction matrix;

[0028] Construct a one-dimensional zero matrix;

[0029] Arrange the data of the initial dimension reduction matrix in order in the one-dimensional zero matrix to obtain a one-dimensional vector.

[0030] In an achievable implementation, a target feature vector is obtained by linearly combining the second matrix and the preset bias vector with the one-dimensional vector, including:

[0031] Multiply the one-dimensional vector by the second matrix to obtain the third matrix;

[0032] The third matrix is ​​added to the preset bias vector to obtain the target feature vector.

[0033] In an achievable implementation, after sending the satisfaction survey information to the user through the information sending channel corresponding to the target probability information, the method further includes:

[0034] Obtain user satisfaction feedback;

[0035] The number of user satisfaction feedback information that is consistent with the preset label information is divided by the total number of user satisfaction feedback information to obtain the target accuracy information;

[0036] According to the target accuracy information, the structure and parameters of the target model are updated.

[0037] In an achievable implementation, before inputting the user data information into the target model and multiplying the user data information by the first matrix using the target model to obtain a feature matrix corresponding to the user data information, the method further includes:

[0038] Get historical user data information;

[0039] Inputting historical user data information into an initial model, and using the initial model to multiply the historical user data information by a first training matrix to obtain a training feature matrix corresponding to the historical user data information;

[0040] Reduce the dimension of the training feature matrix to obtain a training one-dimensional vector;

[0041] A training target feature vector is obtained by linearly combining the second training matrix and the preset training bias vector with the training one-dimensional vector, wherein the second training matrix corresponds to an information sending channel;

[0042] Based on the correspondence between the feature vector and the probability information, a plurality of training probability information to be sent through the information sending channel is determined according to the training target feature vector;

[0043] Calculate the difference between the training label information corresponding to the plurality of training probability information and the preset expected information;

[0044] Update the parameters of the initial model according to the difference between the training label information corresponding to the multiple training probability information and the preset expected information;

[0045] When the difference between the training label information corresponding to the plurality of training probability information and the preset expected information is less than a preset threshold, the initial model is determined to be the target model.

[0046] In a second aspect, an embodiment of the present disclosure provides a device for determining an information sending channel, the device comprising:

[0047] The acquisition module is used to acquire user data information, which includes satisfaction evaluation channel information, satisfaction evaluation status information, satisfaction survey authorization information and complaint information;

[0048] An input module, used to input the user data information into the target model, and use the target model to multiply the user data information with the first matrix to obtain a feature matrix corresponding to the user data information;

[0049] Dimensionality reduction module, used to reduce the dimension of feature matrix to obtain a one-dimensional vector;

[0050] A combination module, used for performing a linear combination of a second matrix and a preset bias vector with the one-dimensional vector to obtain a target feature vector, wherein the second matrix corresponds to an information sending channel;

[0051] A determination module, used for determining a plurality of probability information to be sent through an information sending channel according to a target feature vector based on a correspondence between the feature vector and the probability information;

[0052] A selection module is used to select the maximum value among multiple probability information as target probability information;

[0053] The sending module is used to send satisfaction survey information to users through the information sending channel corresponding to the target probability information.

[0054] In a third aspect, an embodiment of the present disclosure provides a device for determining an information sending channel, the device comprising a processor and a memory storing computer program instructions; the processor reads and executes the computer program instructions to implement a method for determining an information sending channel as described in any one of the first aspects.

[0055] In a fourth aspect, an embodiment of the present disclosure provides a computer-readable storage medium, which stores computer program instructions. When the computer program instructions are executed by a processor, a method for determining an information sending channel as described in any one of the first aspects is implemented.

[0056] In a fifth aspect, an embodiment of the present disclosure provides a computer program product, including a computer program, which, when executed by a processor, implements the method for determining an information sending channel as described in any one of the first aspects.

[0057] The method, device, equipment, computer storage medium and computer program product for determining the information transmission channel provided by the embodiment of the present disclosure obtain user data information, input the user data information into the target model, and use the target model to multiply the user data information with the first matrix to obtain a feature matrix corresponding to the user data information. The feature matrix is ​​reduced in dimension to obtain a one-dimensional vector. By reducing the dimension of the feature matrix, the most important feature information in the user data information can be obtained. The second matrix and the preset bias vector are linearly combined with the one-dimensional vector to obtain the target feature vector, so that the feature data of the user data information establishes a corresponding relationship with the information transmission channel. Based on the corresponding relationship between the feature vector and the probability information, multiple probability information sent through the information transmission channel is determined according to the target feature vector, the maximum value of the multiple probability information is selected as the target probability information, and the satisfaction survey information is sent to the user through the information transmission channel corresponding to the target probability information. The present disclosure extracts the feature matrix of the user data, and reduces the dimension and transforms it based on the feature matrix, so that the information transmission channel information corresponds to the characteristics of the user data information, thereby improving the matching degree between the information transmission channel and the user. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] In order to more clearly illustrate the technical solution of the embodiment of the present invention, the following is a brief introduction to the drawings required for use in the embodiment of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0059] Figure 1 It is a flowchart of a method for determining an information sending channel provided by an embodiment of the present disclosure;

[0060] Figure 2 is a flowchart of a method for obtaining user data information provided by an embodiment of the present disclosure;

[0061] Figure 3 is a flowchart of a method for obtaining a feature matrix provided by an embodiment of the present disclosure;

[0062] Figure 4 is a flowchart of a method for obtaining a one-dimensional vector provided by an embodiment of the present disclosure;

[0063] Figure 5 is a flow chart of a method for obtaining a target feature vector provided by an embodiment of the present disclosure;

[0064] Figure 6 It is a flowchart of another method for determining an information sending channel provided by an embodiment of the present disclosure;

[0065] Figure 7 It is a flowchart of another method for determining an information sending channel provided by an embodiment of the present disclosure;

[0066] Figure 8 It is a structural diagram of a device for determining an information sending channel provided by an embodiment of the present disclosure;

[0067] Fig. 9 It is a structural diagram of a device for determining an information sending channel provided by an embodiment of the present disclosure. DETAILED DESCRIPTION

[0068] The features and exemplary embodiments of various aspects of the present invention will be described in detail below. In order to make the purpose, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only intended to explain the present invention, rather than to limit the present invention. For those skilled in the art, the present invention can be implemented without the need for some of these specific details. The following description of the embodiments is only to provide a better understanding of the present invention by illustrating examples of the present invention.

[0069] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the statement "include..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0070] It should be noted that the acquisition, storage, use and processing of data in the embodiments of the present application are in compliance with the relevant provisions of national laws and regulations.

[0071] It should be noted that in the embodiments of the present application, certain software, components, models and other existing solutions in the industry may be mentioned, and they should be regarded as exemplary. Their purpose is only to illustrate the feasibility of implementing the technical solution of the present application, but it does not mean that the applicant has or will necessarily use the solution.

[0072] Before describing the technical solutions provided by the embodiments of the present disclosure, in order to facilitate understanding of the embodiments of the present disclosure, the present disclosure specifically describes the problems existing in the related technologies:

[0073] At present, with the continuous competition in the service industry, more and more companies are taking various measures to improve service quality in order to win the favor of users, such as providing caring services to customers, providing precision marketing services to customers, or conducting satisfaction surveys on users.

[0074] Among them, the specific method of conducting a satisfaction survey on users is: after the user completes the business, the company sends satisfaction survey information to the user through the background server to improve service quality based on the user's feedback information.

[0075] However, in the prior art, satisfaction survey information is often sent to users in a single survey method, such as interactive voice response (IVR), text messages, WeChat, etc. A single survey method cannot cover the preferences of all users. For example, some users are disgusted with sending text messages or WeChat, and some users are unwilling to answer interactive voice calls. This results in a low match between the information sending channel and the user, resulting in a low evaluation rate of users participating in the satisfaction survey, and the company is unable to make targeted service improvements for users.

[0076] The embodiments of the present disclosure provide a method, apparatus, device, computer storage medium and computer program product for determining an information sending channel, which can solve the technical problems existing in the related technologies.

[0077] The disclosed embodiment obtains user data information, inputs the user data information into the target model, and uses the target model to multiply the user data information with the first matrix to obtain a feature matrix corresponding to the user data information. The feature matrix is ​​reduced in dimension to obtain a one-dimensional vector. By reducing the dimension of the feature matrix, the most important feature information in the user data information can be obtained. The second matrix and the preset bias vector are linearly combined with the one-dimensional vector to obtain the target feature vector, so that the feature data of the user data information establishes a corresponding relationship with the information sending channel. Based on the corresponding relationship between the feature vector and the probability information, multiple probability information sent through the information sending channel is determined according to the target feature vector, and the maximum value of the multiple probability information is selected as the target probability information, and the satisfaction survey information is sent to the user through the information sending channel corresponding to the target probability information. The disclosed embodiment extracts the feature matrix of the user data, and reduces and transforms the dimension based on the feature matrix, so that the information sending channel information corresponds to the characteristics of the user data information, thereby improving the matching degree between the information sending channel and the user.

[0078] The following is an introduction to the method for determining the information sending channel provided by the embodiment of the present disclosure.

[0079] Figure 1 FIG. 1 is a flow chart showing a method for determining an information transmission channel provided by an embodiment of the present disclosure. Figure 1, the method includes steps S110-S170.

[0080] S110: Obtain user data information.

[0081] Use the data acquisition module to obtain user data information.

[0082] Among them, user data information includes satisfaction evaluation channel information, satisfaction evaluation status information, satisfaction survey authorization information and complaint information. User data information can be represented in the form of a matrix. Satisfaction evaluation channel information represents the channel used by users to evaluate and feedback the satisfaction survey. Satisfaction evaluation status information represents the score of users' evaluation and feedback on the satisfaction survey. Complaint information includes complaint intention information, and complaint intention information represents the user's complaint tendency to the enterprise's satisfaction survey.

[0083] In one example, user data information may be stored in components related to the big data platform.

[0084] In one example, the user data information may also include customer star rating information and peace of mind contract status information. Customer star rating information represents the service level enjoyed by the customer in the enterprise, and the customer star rating information can be used to filter user data information. Peace of mind contract status information represents the user's authorization information to accept the enterprise to send business-related information.

[0085] The disclosed embodiment acquires user data information to facilitate subsequent analysis of the user data information and establish a corresponding relationship between the user data information and the information sending channel.

[0086] S120: Input the user data information into the target model, and use the target model to multiply the user data information by the first matrix to obtain a feature matrix corresponding to the user data information.

[0087] After the user data information is acquired, the user data information is input into the target model. The preset first matrix is ​​multiplied by the target model to obtain a feature matrix corresponding to the user data information.

[0088] In one example, the target model may be a convolutional neural network (CNN) or a recurrent neural network (RSN) or a LSTM neural network (LSTM).

[0089] In one example, the preset first matrix may be a convolution kernel.

[0090] S130: Dimensionality reduction processing is performed on the feature matrix to obtain a one-dimensional vector.

[0091] After obtaining the feature matrix, the dimension of the feature matrix is ​​reduced to obtain a one-dimensional vector.

[0092] In one example, the dimensionality reduction process may be performed by multiplying the feature matrix by a matrix having a smaller number of rows and columns than the feature matrix.

[0093] S140: Use the second matrix and the preset bias vector to perform a linear combination with the one-dimensional vector to obtain a target feature vector.

[0094] After the feature matrix is ​​reduced in dimension to obtain a one-dimensional vector, the second matrix and the preset bias vector are linearly combined with the one-dimensional vector to obtain a target feature vector.

[0095] The second matrix corresponds to the information sending channel, that is, the second matrix represents the weight matrix of the information sending channel.

[0096] S150: Based on the correspondence between the feature vector and the probability information, determine a plurality of probability information to be sent through the information sending channel according to the target feature vector.

[0097] After the target feature vector is obtained, the corresponding relationship between the feature vector and the probability information is used to determine the multiple probability information to be sent through the information sending channel according to the target feature vector.

[0098] In one example, the correspondence between the feature vector and the probability information may be a softmax activation function.

[0099] In another example, based on the correspondence between the feature vector and the probability information, the steps of determining multiple probability information sent through the information sending channel according to the target feature vector are as follows: wherein the target feature vector is [1 2 3]. Using the softmax activation function, three probability information are determined according to the target feature vector [1 2 3]. wherein the first probability information is The second probability information is The third probability information is

[0100] S160: Selecting the maximum value among the multiple probability information as the target probability information.

[0101] After obtaining multiple probability information, the values ​​of the multiple probability information are compared, and the maximum value among the multiple probability information is selected as the target probability information.

[0102] In one example, multiple probability information can be obtained through the target model, for example, the probability of interactive voice is 60%, the probability of WeChat is 20%, and the probability of SMS is 20%. Then the probability information of interactive voice, that is, 60%, is selected as the target probability information.

[0103] S170: Sending satisfaction survey information to the user through an information sending channel corresponding to the target probability information.

[0104] After determining the target probability information, the satisfaction survey information is sent to the user through the information sending channel corresponding to the target probability information.

[0105] In one example, the probability of interactive voice is 60%, the probability of WeChat is 20%, and the probability of SMS is 20% obtained through the target model. Then the probability information of interactive voice, i.e. 60%, is selected as the target probability information, and the satisfaction survey information is sent to the user through the interactive voice information sending channel.

[0106] The disclosed embodiment extracts a feature matrix of user data and performs dimensionality reduction and transformation based on the feature matrix, so that the information sending channel information corresponds to the features of the user data information, thereby improving the matching degree between the information sending channel and the user.

[0107] In one embodiment, Figure 2 As shown, obtaining user data information may include steps S111-S114.

[0108] S111: Acquire initial user data information.

[0109] Use the data acquisition module to obtain initial user data information.

[0110] Among them, the initial user data information may include satisfaction evaluation channel information, satisfaction evaluation status information, satisfaction survey authorization information, complaint information, customer star information and peace of mind contract status information.

[0111] S112: Delete missing values, duplicate values, and abnormal values ​​of the initial user data information to obtain first user data information.

[0112] After the initial user data information is acquired, data preprocessing is performed on the initial user data information to detect missing values, duplicate values ​​and abnormal values ​​in the initial user data information and delete them to obtain the first user data information.

[0113] In one example, after detecting missing values ​​and abnormal values ​​in the initial user data information, the missing values ​​in the initial user data information can be filled, for example, by filling with default values ​​such as 0, mean or median, or filling based on difference method or filling based on output results of the model. The abnormal values ​​in the initial user data information can also be modified, for example, by replacing them with normal values ​​or correcting them based on statistical methods.

[0114] In another example, features that have a greater impact on customer satisfaction in the first user data information may be selected based on prior knowledge for subsequent processing.

[0115] S113: Convert the non-numeric data in the first user data information into numeric data to obtain second user data information.

[0116] After the first user data information is acquired, the non-numeric data in the first user data information is converted into numeric data using a preset conversion method to obtain the second user data information.

[0117] In one example, the satisfaction evaluation status information can be converted as follows: 1 means very satisfied, 2 means satisfied, 3 means dissatisfied, 4 means dissatisfied other, and 5 means very dissatisfied. The customer star rating information can be converted as follows: 1 means one star, 2 means two stars, 3 means three stars, 4 means four stars, and 5 means five stars. The satisfaction evaluation channel information can be converted as follows: 1 means no evaluation, 2 means interactive voice, 3 means text message, 4 means WeChat, and 5 means manual channel. The peace of mind right contract status information can be converted as follows: 0 means the user has authorized, and 1 means the user has not authorized. The complaint information can be converted as follows: 0 means the user has no complaint tendency, 1 means the user has a slight complaint intention, 2 means the user is very likely to complain, and 3 means the user has already complained.

[0118] The embodiment of the present disclosure converts non-numeric data in the first user data information into numeric data, so that the format of the converted data meets the input requirements of the target model.

[0119] S114: normalize the second user data information to obtain user data information.

[0120] After the second user data information is acquired, the second user data information is normalized using a preset encoding method to obtain user data information.

[0121] In one example, the preset encoding method may include using a one-hot encoding method.

[0122] The embodiment of the present disclosure deletes, transforms and normalizes the initial user data information so that the processed data can meet the input requirements of the target model, reduce noise and redundant information, and improve the data quality of the input data.

[0123] In one embodiment, Figure 3 As shown, multiplying the user data information by the first matrix to obtain a feature matrix corresponding to the user data information may include S121-S124.

[0124] S121: Construct a first zero matrix.

[0125] Construct a first zero matrix.

[0126] The number of rows of the first zero matrix is ​​equal to the number of rows of the first matrix, and the number of columns of the first zero matrix is ​​equal to the number of columns of the first matrix. The element data of each row and each column of the first zero matrix is ​​0.

[0127] In an example, the first matrix is ​​a matrix with 3 rows and 3 columns, and the first zero matrix is ​​a zero matrix with 3 rows and 3 columns.

[0128] S122: Divide the user data information into multiple user data sub-modules.

[0129] In one embodiment, the matrix corresponding to the user data information is divided into a plurality of user data sub-modules, namely user data sub-matrices, according to a preset division method.

[0130] The number of rows of the user data submatrix is ​​equal to the number of rows of the first matrix, and the number of columns of the user data submatrix is ​​equal to the number of columns of the first matrix.

[0131] In an example, the preset division method may be a division method from top to bottom, from left to right, with a step size of 1.

[0132] S123: multiplying the first matrix by a plurality of user data sub-modules respectively according to a first preset order to obtain a plurality of sub-data.

[0133] After the user data information is divided into a plurality of user data sub-modules, the first matrix is ​​multiplied by the plurality of user data sub-modules respectively according to a first preset order to obtain a plurality of sub-data.

[0134] In an example, the first preset order may be an order from top to bottom, from left to right, with a step length of 1.

[0135] S124: Input the plurality of sub-data into the first zero matrix according to a first preset order to obtain a characteristic matrix.

[0136] After obtaining the plurality of sub-data, the plurality of sub-data are sequentially input into the first zero matrix according to a first preset order to obtain a characteristic matrix.

[0137] In one example, the first zero matrix is The matrix corresponding to the user data information is: The matrix corresponding to the user data information is divided into user data sub-matrices in sequence from top to bottom and from left to right with a step size of 1. User Data Submatrix User Data Submatrix User Data Submatrix User Data Submatrix User Data Submatrix User Data Submatrix User Data Submatrix User Data Submatrix Using the first matrix The above 9 user data sub-matrices are multiplied in a preset order to obtain 9 sub-data: A1, A2, A3, B1, B2, B3, C1, C2, C3. The 9 sub-data are input into the first zero matrix in a preset order to obtain the characteristic matrix

[0138] The embodiment of the present disclosure can obtain important characteristic data of the user data information by multiplying the preset first matrix with the user data information, so as to facilitate the subsequent establishment of a corresponding relationship between the user data information and the information sending channel according to the characteristic data.

[0139] In one embodiment, Figure 4 As shown, the dimension reduction process of the feature matrix to obtain a one-dimensional vector may include steps S131-S136.

[0140] S131: Construct a second zero matrix.

[0141] Construct a second zero matrix.

[0142] The number of rows of the second zero matrix is ​​less than the number of rows of the characteristic matrix, the number of columns of the second zero matrix is ​​less than the number of columns of the characteristic matrix, and the element data of each row and column of the second zero matrix is ​​0.

[0143] In an example, the characteristic matrix is ​​a matrix with 3 rows and 3 columns, and the second zero matrix may be a zero matrix with 2 rows and 2 columns.

[0144] S132: Divide the feature matrix into multiple feature sub-modules.

[0145] In one embodiment, the feature matrix is ​​divided into a plurality of feature sub-modules, namely, feature sub-matrices, according to a preset division method.

[0146] In an example, the preset division method may be a division method from top to bottom, from left to right, with a step size of 1.

[0147] S133: Selecting maximum values ​​from a plurality of feature submodules respectively according to a second preset order to obtain a plurality of feature maximum values.

[0148] After acquiring a plurality of characteristic submodules, the maximum values ​​of the plurality of characteristic submodules are respectively selected according to a second preset order to obtain a plurality of characteristic maximum values.

[0149] In an example, the second preset order may be a division method from top to bottom, from left to right, with a step size of 1.

[0150] S134: Inputting a plurality of feature maximum values ​​into a second zero matrix according to a second preset order to obtain an initial dimension reduction matrix.

[0151] After obtaining a plurality of feature maximum values, the obtained plurality of feature maximum values ​​are input into a second zero matrix according to a second preset order to obtain an initial dimension reduction matrix.

[0152] S135: Construct a one-dimensional zero matrix.

[0153] Construct a one-dimensional zero matrix.

[0154] Among them, the number of rows and columns of the one-dimensional zero matrix is ​​1. The element data of the one-dimensional zero matrix are all 0. The number of elements of the one-dimensional zero matrix is ​​equal to the number of elements of the initial dimension reduction matrix.

[0155] S136: Arrange the data of the initial dimension reduction matrix in order in a one-dimensional zero matrix to obtain a one-dimensional vector.

[0156] Each row of data of the obtained initial dimension reduction matrix is ​​arranged in order on the one-dimensional zero matrix to obtain a one-dimensional vector.

[0157] In one example, the second zero matrix is The feature matrix is Divide the feature matrix into feature sub-matrices in sequence from top to bottom, from left to right, with a step size of 1 Eigensubmatrix Eigensubmatrix and the eigenmatrix Select the maximum values ​​of the above multiple feature submodules in the order from top to bottom, from left to right, and step length 1, and obtain multiple feature maximum values: B2, B3, C2, C3. Input the obtained multiple feature maximum values ​​into the second zero matrix in the order from top to bottom, from left to right, and step length 1 to obtain the initial dimension reduction matrix Construct a one-dimensional zero matrix [0 0 0 0]. Arrange each row of the obtained initial dimension reduction matrix in order on the one-dimensional zero matrix to obtain a one-dimensional vector [B2 B3 C2 C3].

[0158] The disclosed embodiment reduces the dimension of the feature matrix to obtain a one-dimensional vector, which facilitates the subsequent determination of the target feature vector based on the one-dimensional vector.

[0159] In one embodiment, Figure 5 As shown, the target feature vector is obtained by linearly combining the second matrix and the preset bias vector with the one-dimensional vector, which may include step S141 and step S142.

[0160] S141: Multiply the one-dimensional vector by the second matrix to obtain a third matrix.

[0161] After obtaining the one-dimensional vector, the one-dimensional vector is multiplied by the second matrix to obtain the third matrix.

[0162] The number of columns of the one-dimensional vector is equal to the number of rows of the second matrix.

[0163] In one example, the one-dimensional vector [B2 B3 C2 C3], the second matrix is Then the one-dimensional vector is multiplied by the second matrix to obtain the third matrix [h1 h2 h3].

[0164] S142: Add the third matrix to the preset bias vector to obtain a target feature vector.

[0165] After the third matrix is ​​obtained, the third matrix is ​​added to a preset bias vector to obtain a target feature vector.

[0166] In an example, the third matrix is ​​[h1 h2 h3], and the third matrix is ​​added to a preset bias vector [l1 l2 l3] to obtain a target feature vector [n1 n2 n3].

[0167] The disclosed embodiment utilizes a second matrix and a preset bias vector to linearly combine with a one-dimensional vector to obtain a target feature vector, which can make the target feature vector fuse the features of the feature data obtained previously, making it more representative and discriminative than the feature data obtained previously.

[0168] In one embodiment, Figure 6 As shown, after sending the satisfaction survey information to the user through the information sending channel corresponding to the target probability information, the method for determining the information sending channel further includes steps S210-S230.

[0169] S210: Obtain user satisfaction feedback information.

[0170] Get satisfaction feedback information sent by users.

[0171] Among them, the satisfaction feedback information includes satisfaction evaluation status information.

[0172] S220: Divide the number of user satisfaction feedback information that is consistent with the preset label information by the total number of user satisfaction feedback information to obtain target accuracy information.

[0173] After obtaining the user satisfaction feedback information, the number of user satisfaction feedback information that is consistent with the preset label information is divided by the total number of user satisfaction feedback information to obtain the target accuracy information.

[0174] The preset label information is determined based on the output result of the target model.

[0175] S230: Update the structure and parameters of the target model according to the target accuracy information.

[0176] After the target accuracy information is calculated, the structure and parameters of the target model are updated according to the target accuracy information.

[0177] In one example, the number of intermediate layers of the target model can be increased according to the target accuracy information, or the number of rows and columns of the first matrix can be adjusted, or data enhancement processing can be performed on the input data of the target model.

[0178] The disclosed embodiment determines the target accuracy information based on the user satisfaction feedback information, and can adjust the structure and parameters of the target model according to the target accuracy information, thereby improving the accuracy of the output result of the target model.

[0179] In one embodiment, Figure 7 As shown, before inputting the user data information into the target model and multiplying the user data information with the first matrix using the target model to obtain a feature matrix corresponding to the user data information, the method for determining the information sending channel also includes steps S310-S380.

[0180] S310: Obtain historical user data information.

[0181] Use the data collection module to obtain historical user data information.

[0182] Among them, historical user data information includes historical satisfaction evaluation channel information, historical satisfaction evaluation status information, historical satisfaction survey authorization information and historical complaint information.

[0183] In one example, historical user data information may be stored in components related to the big data platform.

[0184] S320: Input the historical user data information into the initial model, and use the initial model to multiply the historical user data information with the first training matrix to obtain a training feature matrix corresponding to the historical user data information.

[0185] After the historical user data information is obtained, the historical user data information is input into the initial model. The initial model multiplies the preset first training matrix with the historical user data information to obtain a training feature matrix corresponding to the historical user data information.

[0186] In one example, the first training matrix may be a convolution kernel.

[0187] S330: Perform dimensionality reduction processing on the training feature matrix to obtain a training one-dimensional vector.

[0188] After obtaining the training feature matrix, the dimension of the training feature matrix is ​​reduced to obtain a training one-dimensional vector.

[0189] In one example, the dimensionality reduction process may be performed by multiplying the training feature matrix by a matrix having a smaller number of rows and columns than the training feature matrix.

[0190] S340: Use the second training matrix and the preset training offset vector to perform linear combination with the training one-dimensional vector to obtain a training target feature vector.

[0191] After the training feature matrix is ​​reduced in dimension to obtain a training one-dimensional vector, a second training matrix and a preset training bias vector are linearly combined with the training one-dimensional vector to obtain a training target feature vector.

[0192] The second training matrix corresponds to the information sending channel.

[0193] S350: Based on the correspondence between the feature vector and the probability information, determine a plurality of training probability information to be sent through the information sending channel according to the training target feature vector.

[0194] After the training target feature vector is obtained, a plurality of training probability information to be sent through an information sending channel is determined according to the training target feature vector through the correspondence between the feature vector and the probability information.

[0195] S360: Calculate the difference between the training label information corresponding to the multiple training probability information and the preset expected information.

[0196] After obtaining multiple training probability information, corresponding training label information is determined according to the multiple training probability information, and the difference between the training label information and the preset expected information is calculated.

[0197] In one example, a loss function may be used to calculate the difference between training label information and preset expected information.

[0198] S370: Update the parameters of the initial model according to the difference between the training label information corresponding to the multiple training probability information and the preset expected information.

[0199] In one embodiment, after calculating the difference between the training label information corresponding to the plurality of training probability information and the preset expected information, the gradient of the difference relative to the parameters in the target model can be calculated using a back propagation algorithm, and the parameters in the target model are updated in a gradient descent manner.

[0200] S380: When the difference between the training label information corresponding to the plurality of training probability information and the preset expected information is less than a preset threshold, determining the initial model as the target model.

[0201] When the parameters of the target model are continuously iterated, if the difference between the training label information corresponding to the multiple training probability information and the preset expected information is less than a preset threshold, the initial model is determined to be the target model.

[0202] The disclosed embodiment updates the parameters of the initial model by the difference between the output result of the initial model and the preset expected information, which can improve the accuracy of the output result of the target model.

[0203] like Figure 8 As shown, the embodiment of the present disclosure further provides a device 800 for determining an information sending channel, and the device 800 for determining an information sending channel includes:

[0204] The acquisition module 801 is used to acquire user data information, which includes satisfaction evaluation channel information, satisfaction evaluation status information, satisfaction survey authorization information and complaint information;

[0205] An input module 802 is used to input the user data information into the target model, and use the target model to multiply the user data information with the first matrix to obtain a feature matrix corresponding to the user data information;

[0206] A dimension reduction module 803 is used to reduce the dimension of the feature matrix to obtain a one-dimensional vector;

[0207] A combination module 804 is used to use the second matrix and the preset bias vector to perform a linear combination with the one-dimensional vector to obtain a target feature vector, wherein the second matrix corresponds to an information sending channel;

[0208] A determination module 805, configured to determine a plurality of probability information to be sent through an information sending channel according to a target feature vector based on a correspondence between the feature vector and the probability information;

[0209] A selection module 806 is used to select a maximum value among the plurality of probability information as target probability information;

[0210] The sending module 807 is used to send the satisfaction survey information to the user through the information sending channel corresponding to the target probability information.

[0211] The disclosed embodiment extracts a feature matrix of user data and performs dimensionality reduction and transformation based on the feature matrix, so that the information sending channel information corresponds to the features of the user data information, thereby improving the matching degree between the information sending channel and the user.

[0212] In some embodiments, the acquisition module 801 is specifically used to:

[0213] Get initial user data information;

[0214] Deleting missing values, duplicate values, and abnormal values ​​of the initial user data information to obtain first user data information;

[0215] Converting non-numeric data in the first user data information into numeric data to obtain second user data information;

[0216] The second user data information is normalized to obtain user data information.

[0217] In some embodiments, the information transmission channel determination device 800 further includes a construction module, the construction module is used to construct a first zero matrix, the number of rows of the first zero matrix is ​​equal to the number of rows of the first matrix, and the number of columns of the first zero matrix is ​​equal to the number of columns of the first matrix;

[0218] Dividing the user data information into a plurality of user data submodules;

[0219] Multiplying the first matrix by the plurality of user data submodules respectively according to a first preset order to obtain a plurality of sub-data;

[0220] A plurality of sub-data are input into a first zero matrix according to a first preset order to obtain a characteristic matrix.

[0221] In some embodiments, the information transmission channel determination device 800 further includes a construction module, the construction module is used to construct a second zero matrix, the number of rows of the second zero matrix is ​​less than the number of rows of the characteristic matrix, and the number of columns of the second zero matrix is ​​less than the number of columns of the characteristic matrix;

[0222] Divide the feature matrix into multiple feature submodules;

[0223] Selecting maximum values ​​of multiple feature submodules respectively according to a second preset order to obtain multiple feature maximum values;

[0224] Inputting a plurality of feature maximum values ​​into a second zero matrix according to a second preset order to obtain an initial dimension reduction matrix;

[0225] Construct a one-dimensional zero matrix;

[0226] Arrange the data of the initial dimension reduction matrix in order in the one-dimensional zero matrix to obtain a one-dimensional vector.

[0227] In some embodiments, the device 800 for determining an information sending channel further includes a calculation module, which is used to multiply the one-dimensional vector by the second matrix to obtain a third matrix; and add the third matrix to a preset bias vector to obtain a target feature vector.

[0228] In some embodiments, the acquisition module 801 is specifically used to:

[0229] Obtain user satisfaction feedback;

[0230] The number of user satisfaction feedback information that is consistent with the preset label information is divided by the total number of user satisfaction feedback information to obtain the target accuracy information;

[0231] According to the target accuracy information, the structure and parameters of the target model are updated.

[0232] In some embodiments, the acquisition module 801 is specifically used to:

[0233] Get historical user data information;

[0234] Inputting historical user data information into an initial model, and using the initial model to multiply the historical user data information by a first training matrix to obtain a training feature matrix corresponding to the historical user data information;

[0235] Reduce the dimension of the training feature matrix to obtain a training one-dimensional vector;

[0236] A training target feature vector is obtained by linearly combining the second training matrix and the preset training bias vector with the training one-dimensional vector, wherein the second training matrix corresponds to an information sending channel;

[0237] Based on the correspondence between the feature vector and the probability information, a plurality of training probability information to be sent through the information sending channel is determined according to the training target feature vector;

[0238] Calculate the difference between the training label information corresponding to the plurality of training probability information and the preset expected information;

[0239] Update the parameters of the initial model according to the difference between the training label information corresponding to the multiple training probability information and the preset expected information;

[0240] When the difference between the training label information corresponding to the plurality of training probability information and the preset expected information is less than a preset threshold, the initial model is determined to be the target model.

[0241] Figure 8 Each module / unit in the device shown has the function of realizing Figures 1 to 7 The functions of each step executed by the application system in the present invention can achieve the corresponding technical effects. For the sake of brevity, they will not be repeated here.

[0242] Fig. 9 A schematic diagram of the hardware structure for determining the information sending channel provided by an embodiment of the present disclosure is shown.

[0243] The device determined in the information transmission channel may include a processor 901 and a memory 902 storing computer program instructions.

[0244] Specifically, the processor 901 may include a central processing unit (CPU), or an application specific integrated circuit (ASIC), or may be configured to implement one or more integrated circuits of the embodiments of the present application.

[0245] The memory 902 may include a large capacity memory for data or instructions. By way of example and not limitation, the memory 902 may include a hard disk drive (HDD), a floppy disk drive, a flash memory, an optical disk, a magneto-optical disk, a magnetic tape, or a universal serial bus (USB) drive or a combination of two or more of these. In appropriate cases, the memory 902 may include a removable or non-removable (or fixed) medium. In appropriate cases, the memory 902 may be inside or outside the integrated gateway disaster recovery device. In a specific embodiment, the memory 902 is a non-volatile solid-state memory.

[0246] The memory 902 may include a read-only memory (ROM), a random access memory (RAM), a magnetic disk storage medium device, an optical storage medium device, a flash memory device, an electrical, optical or other physical / tangible memory storage device. Therefore, generally, the memory includes one or more tangible (non-transitory) computer-readable storage media (e.g., memory devices) encoded with software including computer-executable instructions, and when the software is executed (e.g., by one or more processors), it is operable to perform the operations described with reference to the method according to one aspect of the present disclosure.

[0247] The processor 901 implements any one of the methods for determining an information sending channel in the above embodiments by reading and executing computer program instructions stored in the memory 902 .

[0248] In one example, the device for determining the information transmission channel may further include a communication interface 903 and a bus 904. Fig. 9 As shown, the processor 901, the memory 902, and the communication interface 903 are connected via a bus 904 and communicate with each other.

[0249] The communication interface 903 is mainly used to implement communication between various modules, devices, units and / or equipment in the embodiments of the present application.

[0250] Bus 904 includes hardware, software or both, and couples the components of the online data traffic billing device to each other. For example, but not limitation, the bus may include an accelerated graphics port (Accelerated Graphics Port, AGP) or other graphics bus, an enhanced industry standard architecture (Extended Industry Standard Architecture, EISA) bus, a front side bus (Front Side Bus, FSB), a hypertransport (Hyper Transport, HT) interconnection, an industry standard architecture (Industry Standard Architecture, ISA) bus, an infinite bandwidth interconnection, a low pin count (LPC) bus, a memory bus, a micro channel architecture (MCA) bus, a peripheral component interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a serial advanced technology attachment (SATA) bus, a video electronics standard association local (VLB) bus or other suitable buses or a combination of two or more of these. In appropriate cases, bus 904 may include one or more buses. Although the embodiment of the present invention describes and illustrates a specific bus, the present invention considers any suitable bus or interconnection. In addition, in conjunction with the method for determining the information transmission channel in the above embodiment, the embodiment of the present application also provides a computer storage medium to implement. The computer storage medium stores computer program instructions; when the computer program instructions are executed by the processor, any one of the methods for determining an information sending channel in the above-mentioned embodiments is implemented.

[0251] An embodiment of the present application also provides a computer program product, including a computer program, which, when executed by a processor, implements any one of the methods for determining an information sending channel in the above embodiments.

[0252] It should be clear that the present invention is not limited to the specific configuration and processing described above and shown in the figures. For the sake of simplicity, a detailed description of the known method is omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of the present invention is not limited to the specific steps described and shown, and those skilled in the art can make various changes, modifications and additions, or change the order between the steps after understanding the spirit of the present invention.

[0253] The functional blocks shown in the above structural block diagram can be implemented as hardware, software, firmware or a combination thereof. When implemented in hardware, it can be, for example, an electronic circuit, an application specific integrated circuit (Application Specific Integrated Circuit, ASIC), appropriate firmware, plug-in, function card, etc. When implemented in software, the elements of the present invention are programs or code segments used to perform the required tasks. The program or code segment can be stored in a machine-readable medium, or transmitted on a transmission medium or communication link by a data signal carried in a carrier. "Machine-readable medium" can include any medium capable of storing or transmitting information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, read-only memory (Read-Only Memory, ROM), flash memory, erasable read-only memory (Erasable ReadOnly Memory, EROM), floppy disks, compact disc read-only memory (Compact Disc Read-Only Memory, CD-ROM), optical discs, hard disks, optical fiber media, radio frequency (Radio Frequency, RF) links, etc. The code segment can be downloaded via a computer network such as the Internet, an intranet, etc.

[0254] It should also be noted that the exemplary embodiments mentioned in the present invention describe some methods or systems based on a series of steps or devices. However, the present invention is not limited to the order of the above steps, that is, the steps can be performed in the order mentioned in the embodiments, or in a different order from the embodiments, or several steps can be performed simultaneously.

[0255] Aspects of the present disclosure are described above with reference to the flowchart and / or block diagram of the method, device (system) and computer program product according to the embodiment of the present disclosure. It should be understood that each box in the flowchart and / or block diagram and the combination of each box in the flowchart and / or block diagram can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device to produce a machine so that these instructions executed by the processor of the computer or other programmable data processing device enable the implementation of the function / action specified in one or more boxes of the flowchart and / or block diagram. Such a processor can be, but is not limited to, a general-purpose processor, a special-purpose processor, a special application processor, or a field programmable logic circuit. It can also be understood that each box in the block diagram and / or flowchart and the combination of boxes in the block diagram and / or flowchart can also be implemented by dedicated hardware that performs a specified function or action, or can be implemented by a combination of dedicated hardware and computer instructions.

[0256] The above is only a specific implementation of the present invention. Those skilled in the art can clearly understand that for the convenience and simplicity of description, the specific working process of the system, module and unit described above can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here. It should be understood that the protection scope of the present invention is not limited to this. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed by the present invention, and these modifications or replacements should be covered within the protection scope of the present invention.

Claims

1. A method for determining an information sending channel, characterized in that: include: Acquire user data information, wherein the user data information includes satisfaction evaluation channel information, satisfaction evaluation status information, satisfaction survey authorization information and complaint information; Inputting the user data information into a target model, and using the target model to multiply the user data information by a first matrix to obtain a feature matrix corresponding to the user data information; Performing dimensionality reduction processing on the feature matrix to obtain a one-dimensional vector; A target feature vector is obtained by linearly combining a second matrix and a preset bias vector with the one-dimensional vector, wherein the second matrix corresponds to an information sending channel; Based on the correspondence between the feature vector and the probability information, determining a plurality of probability information to be sent through the information sending channel according to the target feature vector; Selecting a maximum value among the plurality of probability information as target probability information; The satisfaction survey information is sent to the user through the information sending channel corresponding to the target probability information.

2. The method according to claim 1, characterized in that The obtaining of user data information includes: Get initial user data information; Deleting missing values, duplicate values, and abnormal values ​​of the initial user data information to obtain first user data information; Converting non-numeric data in the first user data information into numeric data to obtain second user data information; The second user data information is normalized to obtain the user data information.

3. The method according to claim 1, characterized in that The step of multiplying the user data information by the first matrix to obtain a feature matrix corresponding to the user data information includes: Constructing a first zero matrix, wherein the number of rows of the first zero matrix is ​​equal to the number of rows of the first matrix, and the number of columns of the first zero matrix is ​​equal to the number of columns of the first matrix; Dividing the user data information into a plurality of user data submodules; Multiplying the first matrix by the plurality of user data submodules respectively according to a first preset order to obtain a plurality of sub-data; The plurality of sub-data are input into the first zero matrix according to the first preset order to obtain the characteristic matrix.

4. The method according to claim 1, characterized in that The dimensionality reduction process of the feature matrix to obtain a one-dimensional vector includes: Constructing a second zero matrix, wherein the number of rows of the second zero matrix is ​​smaller than the number of rows of the characteristic matrix, and the number of columns of the second zero matrix is ​​smaller than the number of columns of the characteristic matrix; Dividing the feature matrix into a plurality of feature submodules; Selecting the maximum values ​​of the plurality of characteristic submodules respectively according to a second preset order to obtain a plurality of characteristic maximum values; Inputting the plurality of feature maximum values ​​into the second zero matrix according to the second preset order to obtain an initial dimension reduction matrix; Construct a one-dimensional zero matrix; The data of the initial dimension reduction matrix are arranged in order in the one-dimensional zero matrix to obtain the one-dimensional vector.

5. The method according to claim 1, characterized in that The method of linearly combining the one-dimensional vector with the second matrix and the preset bias vector to obtain the target feature vector includes: Multiplying the one-dimensional vector by the second matrix to obtain a third matrix; The third matrix is ​​added to the preset bias vector to obtain the target feature vector.

6. The method according to claim 1, characterized in that After sending the satisfaction survey information to the user through the information sending channel corresponding to the target probability information, the method further includes: Obtain user satisfaction feedback; Dividing the number of user satisfaction feedback information that is consistent with the preset label information by the total number of user satisfaction feedback information to obtain target accuracy information; According to the target accuracy information, the structure and parameters of the target model are updated.

7. The method according to claim 1, characterized in that Before inputting the user data information into a target model and multiplying the user data information by a first matrix using the target model to obtain a feature matrix corresponding to the user data information, the method further includes: Get historical user data information; Inputting the historical user data information into an initial model, and using the initial model to multiply the historical user data information by a first training matrix to obtain a training feature matrix corresponding to the historical user data information; Performing dimensionality reduction processing on the training feature matrix to obtain a training one-dimensional vector; A training target feature vector is obtained by linearly combining a second training matrix and a preset training bias vector with the training one-dimensional vector, wherein the second training matrix corresponds to an information sending channel; Based on the correspondence between the feature vector and the probability information, determining a plurality of training probability information to be sent through the information sending channel according to the training target feature vector; Calculate the difference between the training label information corresponding to the plurality of training probability information and the preset expected information; Update the parameters of the initial model according to the difference between the training label information corresponding to the multiple training probability information and the preset expected information; When the difference between the training label information corresponding to the multiple training probability information and the preset expected information is less than a preset threshold, the initial model is determined to be the target model.

8. A device for determining an information transmission channel, characterized in that: The device comprises: An acquisition module, the acquisition module is used to acquire user data information, the user data information includes satisfaction evaluation channel information, satisfaction evaluation status information, satisfaction survey authorization information and complaint information; An input module, the input module is used to input the user data information into a target model, and use the target model to multiply the user data information with a first matrix to obtain a feature matrix corresponding to the user data information; A dimension reduction module, wherein the dimension reduction module is used to reduce the dimension of the feature matrix to obtain a one-dimensional vector; A combination module, the combination module is used to use a second matrix and a preset bias vector to perform a linear combination with the one-dimensional vector to obtain a target feature vector, wherein the second matrix corresponds to an information sending channel; A determination module, the determination module is used to determine a plurality of probability information sent through an information sending channel according to the target feature vector based on a correspondence between the feature vector and the probability information; A selection module, the selection module is used to select a maximum value among the multiple probability information as the target probability information; A sending module is used to send satisfaction survey information to users through an information sending channel corresponding to the target probability information.

9. A device for determining a channel for sending information, characterized in that: The device includes: a processor, and a memory storing computer program instructions; the processor reads and executes the computer program instructions to implement the method for determining an information sending channel as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer program instructions, and when the computer program instructions are executed by a processor, the method for determining an information sending channel according to any one of claims 1 to 7 is implemented.

11. A computer program product, characterized in that It comprises a computer program, which, when executed by a processor, implements the method for determining an information sending channel as described in any one of claims 1 to 7.

Images