Message pushing method and message pushing device based on full life cycle of user

Abstract

The invention provides a message pushing method and a message pushing device based on a full life cycle of a user. The method comprises the following steps: acquiring source data of a to-be-pushed user; analyzing and dividing to-be-pushed users according to the source data, and dividing the to-be-pushed users to corresponding target life cycle stages and target groups; determining a message content and a message template corresponding to the to-be-pushed user according to the target life cycle stage and the source data; a target pushing strategy corresponding to the user to be pushed is selected according to the target life cycle stage and the target group, the target pushing strategy is any pushing strategy, and the pushing strategy comprises a pushing channel, pushing frequency and pushing time; and pushing the message content to the to-be-pushed user according to the target pushing strategy and the message template. According to the method, the problem that in the prior art, full-life-cycle division and classification are not carried out on users, personalized, customized and targeted message pushing cannot be provided, and consequently part of user experience is poor is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of message push technology, and in particular to a message push method based on the entire life cycle of a user, a message push device, a computer-readable storage medium and a computer program product. Background Art

[0002] With the popularity of mobile devices and the development of the Internet, users have an increasing demand for real-time messages. However, the redundancy and overload of messages also bring problems to users. Users expect to receive personalized messages related to their interests and needs. In addition, active users are the basis for the survival of products. Only with users can products bring value. The main goal of message push is to awaken inactive and dormant users, promote their activity, and retain lost users. In the process of using applications, a lot of application behavior data will be generated, such as user behavior (operation behavior, user browsing behavior, user setting behavior, etc.). How to effectively mine these application data to determine relevant user characteristics and needs for subsequent message push is actively explored by most existing social, shopping, reading, and short video software. However, the existing processing methods do not take into account the user's life cycle stage when pushing relevant messages to users, and do not take into account the selection of push strategies such as push channel, push frequency, and push time. Among them, improper selection of push channel is easy for users to ignore information, and excessive push frequency and improper push time are easy to cause users to feel bored, which seriously damages the user's experience. Summary of the invention

[0003] The main purpose of this application is to provide a message push method, a message push device, a computer-readable storage medium and a computer program product based on the entire life cycle of a user, so as to at least solve the problem that the prior art does not divide and classify users throughout their life cycle, fails to provide personalized, customized and targeted message push, and leads to poor user experience for some users.

[0004] In order to achieve the above-mentioned purpose, according to one aspect of the present application, a message push method based on the entire life cycle of a user is provided, including: obtaining source data of the user to be pushed, the source data at least including user basic information, user behavior data, user scenario information, real-time location information, social relationships and message sending records; analyzing and dividing the user to be pushed according to the source data, and dividing the user to be pushed into corresponding target life cycle stages and target groups, the target life cycle stage is any stage in the entire life cycle stage, the entire life cycle stage includes acquisition stage, activation stage, participation and use stage, conversion stage, retention and growth stage and churn stage, the target group is at least one of the following: active users, inactive users and churned users; determining the message content and message template corresponding to the user to be pushed according to the target life cycle stage and the source data; selecting the target push strategy corresponding to the user to be pushed according to the target life cycle stage and the target group, the target push strategy is any push strategy, and the push strategy includes push channel, push frequency and push time; pushing the message content to the user to be pushed according to the target push strategy and the message template.

[0005] Optionally, the users to be pushed are analyzed and divided according to the source data, and the users to be pushed are divided into corresponding target life cycle stages and target groups, including: constructing a target undirected weighted complex network based on mutual information at least according to the source data of the users to be pushed; analyzing the target undirected weighted complex network using a community detection algorithm based on node similarity according to the source data, and dividing all the users to be pushed into the corresponding target life cycle stages and the target groups.

[0006] Optionally, a target undirected weighted network is constructed based on mutual information at least according to the source data of the user to be pushed, including: a selection step, selecting a message push data set, the message push data set is divided into a global data set and a sample data set, the global data set is a collection of all source data of all users and all life cycle stages, and the sample data set is a collection of the source data corresponding to all the users to be pushed in the global data set; a preprocessing step, performing a preprocessing operation on the message push data set to obtain source data features, the preprocessing operation includes a data cleaning operation and a feature selection operation, and the source data features include at least social relationship features, behavior similarity, scenario matching and location similarity; a construction step, defining the user basic information in the message push data set after the preprocessing operation as a node and defining the source data features as an edge and assigning weights, and constructing a target undirected weighted complex network through mutual information, the mutual information is used to measure the degree of association between node pairs, the node pair is any two of the nodes, and one node represents one user to be pushed.

[0007] Optionally, the message push data set is preprocessed to obtain source data features, including: performing the data cleaning operation on the message push data set to remove noise and abnormal data to obtain a valid message push data set; performing feature extraction on the valid message push data set using correlation analysis, and calculating the similarity between the features through correlation coefficients or covariance to obtain the source data features.

[0008] Optionally, constructing a target undirected weighted complex network through mutual information includes: a first calculation step, calculating the mutual information between any two nodes, and constructing a current undirected weighted complex network based on the mutual information; a second calculation step, calculating the characteristic path length and aggregation coefficient of the current undirected weighted complex network, wherein the characteristic path length is the average of the path lengths of all the node pairs in the current undirected weighted complex network, and the aggregation coefficient is used to characterize the overlap between two adjacent nodes; a repeating step, determining the mutual information between any two nodes when the characteristic path length is less than a set path length threshold and / or the aggregation coefficient is less than a aggregation coefficient threshold. The current undirected weighted complex network is unqualified, and the selection step, the preprocessing step, the construction step, the first calculation step, the second calculation step and the repetition step are re-executed in sequence at least once until the characteristic path length is greater than or equal to the set path length threshold and the aggregation coefficient is greater than or equal to the aggregation coefficient threshold; when the characteristic path length is greater than or equal to the set path length threshold and the aggregation coefficient is greater than or equal to the aggregation coefficient threshold, the current undirected weighted complex network is determined to be qualified, and the current undirected weighted complex network is determined as the target undirected weighted complex network.

[0009] Optionally, the target undirected weighted complex network is analyzed according to the source data using a community detection algorithm of node similarity, and all the users to be pushed are divided into the corresponding target life cycle stages and the target groups, including: determining a value range of a similarity threshold according to a target similarity index, and selecting a target similarity threshold within the value range, wherein the target similarity index is at least one of the following: a Salton index, a Sorensen index, and a hub promotion index, and the similarity threshold represents the similarity between any two nodes; an expansion step, adding the neighbor nodes of all neighbor nodes of the current expansion node that meet a first expansion condition to the current community, wherein the current expansion node is any node in the target undirected weighted complex network, and the expansion condition is that the similarity value between the neighbor node and the current expansion node is greater than the target similarity threshold, and the neighbor node is other nodes connected to the node; an updating step, updating the current expansion node in the current community, and repeating the expansion step until there is no neighbor node whose similarity value with the current expansion node is greater than the target similarity threshold, and the neighbor nodes are other nodes connected to the node. The current community is used to characterize any one of the stages of the full life cycle; the target similarity threshold is updated, the expansion step and the update step are sequentially repeated at least once until the set maximum number of repetitions is reached, and remaining nodes and multiple initial communities are obtained, wherein the remaining nodes are nodes that are not divided into any of the initial communities, and the number of the initial communities is less than or equal to the number of stages of the full life cycle; when the number of the remaining nodes is not zero, the remaining nodes are merged into adjacent initial communities according to the similarity index of the remaining nodes, or, when the number of the remaining nodes is not zero and the remaining nodes are not connected to any of the nodes, the remaining nodes are formed into a separate community; the full life cycle stage is determined according to all the initial communities or all the initial communities and the separate communities; user information in each of the communities is analyzed and users in each of the communities are divided into different user groups, wherein the user information at least includes user behavior information and preference information; the community in which the user to be pushed is located is determined as the target life cycle stage, and the user group in which the user to be pushed is located is determined as the target group.

[0010] Optionally, after pushing the message content to the user to be pushed according to the target push strategy and the message template, the method further includes: obtaining user feedback data, the user feedback data at least including the reach rate, click rate and conversion rate of the pushed users, the pushed users being the users after the message is pushed to the user to be pushed; optimizing the message push data set, undirected weighted complex network and community detection algorithm according to the user feedback data; and regularly performing data analysis on the behavior patterns of all the pushed users through the user feedback data to adjust the target push strategy.

[0011] According to another aspect of the present application, a message push device based on the entire life cycle of a user is provided, the device comprising: a first acquisition unit, used to acquire source data of a user to be pushed, the source data comprising at least basic user information, user behavior data, user scenario information, real-time location information, social relationships, and message sending records; a division unit, used to analyze and divide the user to be pushed according to the source data, and divide the user to be pushed into corresponding target life cycle stages and target groups, the target life cycle stage being any stage in the entire life cycle stage, the entire life cycle stage comprising an acquisition stage, an activation stage, a participation and use stage, a transfer stage the target group is at least one of the following: active users, inactive users, retained users and lost users; a determination unit is used to determine the message content and message template corresponding to the user to be pushed according to the target life cycle stage and the source data; a selection unit is used to select the target push strategy corresponding to the user to be pushed according to the target life cycle stage and the target group, the target push strategy is any push strategy, and the push strategy includes a push channel, a push frequency and a push time; a push unit is used to push the message content to the user to be pushed according to the target push strategy and the message template.

[0012] According to another aspect of the present application, a computer-readable storage medium is provided, wherein the computer-readable storage medium includes a stored program, wherein when the program is executed, the device where the computer-readable storage medium is located is controlled to execute any one of the methods described.

[0013] According to another aspect of the present application, a computer program product is provided, comprising computer instructions, wherein when the computer instructions are executed by a processor, any one of the methods described above is implemented.

[0014] Applying the technical solution of the present application, in a message push method based on the entire life cycle of a user, by obtaining the user's source data, specifically including basic information of the user to be pushed, user behavior data, user scenario information, real-time location information, social relationships, and message sending records, the behavior, interests, and characteristics of the user to be pushed are analyzed and divided, and the users are divided into different life cycle stages and groups, and the target life cycle stage and target group of the user to be pushed are determined. According to the divided target life cycle stage and target group, the corresponding message content and message template are set to ensure that the pushed message matches the user's needs and preferences. According to the target life cycle and target group of the user to be pushed, the appropriate push channel, push time, and push frequency are selected to ensure that the message can be accurately and timely delivered to the user to be pushed. The present application solves the problem that the prior art does not divide and classify users throughout their life cycle, fails to provide personalized, customized, and targeted message push, and leads to poor user experience for some users. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 A hardware structure block diagram of a mobile terminal for executing a message push method based on the entire life cycle of a user provided in an embodiment of the present application is shown;

[0016] Figure 2 A schematic diagram of a process of a message push method based on the entire life cycle of a user provided according to an embodiment of the present application is shown;

[0017] Figure 3 A schematic diagram of the structure of a message push system based on the entire life cycle of a user provided according to an embodiment of the present application is shown;

[0018] Figure 4 A structural block diagram of a message push device based on the entire life cycle of a user provided according to an embodiment of the present application is shown.

[0019] The above drawings include the following reference numerals:

[0020] 102, processor; 104, memory; 106, transmission device; 108, input and output devices. DETAILED DESCRIPTION

[0021] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0022] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present application.

[0023] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequential order. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present application described here. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0024] As introduced in the background technology, when the prior art pushes relevant messages to users, the user's life cycle stage is not taken into consideration, and the selection of push strategies such as push channel, push frequency and push time is not taken into consideration. Improper selection of push channel can easily cause users to ignore information, while excessively high push frequency and improper push time can easily cause users to feel bored, seriously damaging the user experience. In order to solve the problem that the prior art does not divide and classify users throughout their life cycle, fails to provide personalized, customized and targeted message push, resulting in poor user experience for some users, the embodiments of the present application provide a message push method based on the user's entire life cycle, a message push device, a computer-readable storage medium and a computer program product.

[0025] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0026] The method embodiments provided in the embodiments of the present application can be executed in a mobile terminal, a computer terminal or a similar computing device. Taking running on a mobile terminal as an example, Figure 1 1 is a hardware structure block diagram of a mobile terminal of a message push method based on the entire life cycle of a user according to an embodiment of the present invention. Figure 1 As shown, the mobile terminal may include one or more ( Figure 1Only one is shown in the figure) a processor 102 (the processor 102 may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA) and a memory 104 for storing data, wherein the mobile terminal may also include a transmission device 106 and an input / output device 108 for communication functions. It can be understood by those skilled in the art that Figure 1 The structure shown is only for illustration and does not limit the structure of the mobile terminal. Figure 1 More or fewer components as shown, or with Figure 1 Different configurations shown.

[0027] The memory 104 can be used to store computer programs, for example, software programs and modules of application software, such as the computer program corresponding to the message push method based on the user's full life cycle in the embodiment of the present invention. The processor 102 executes various functional applications and data processing by running the computer program stored in the memory 104, that is, the above method is implemented. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory 104 may further include a memory remotely arranged relative to the processor 102, and these remote memories can be connected to the mobile terminal via a network. Examples of the above-mentioned network include but are not limited to the Internet, an intranet, a local area network, a mobile communication network, and a combination thereof. The transmission device 106 is used to receive or send data via a network. The above-mentioned specific examples of the network may include a wireless network provided by a communication provider of the mobile terminal. In one example, the transmission device 106 includes a network adapter (Network Interface Controller, referred to as NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device 106 may be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0028] In this embodiment, a message push method based on the entire life cycle of a user is provided and runs on a mobile terminal, a computer terminal or a similar computing device. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although the logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0029] Figure 2 : is a flowchart of a message push method based on the user's entire life cycle according to an embodiment of the present application. Figure 2As shown, the method comprises the following steps:

[0030] Step S201, obtaining source data of the user to be pushed, the source data at least including user basic information, user behavior data, user context information, real-time location information, social relationships and message sending records.

[0031] Specifically, Figure 3 A message push system based on the entire life cycle of a user is proposed according to an embodiment of the present application. The message push system consists of five modules, including a source data acquisition module, a user analysis and segmentation module, a message content and template management module, a push channel and device management module, and a data analysis and optimization module. The source data acquisition module can obtain the source data of all users, including user basic information, user behavior data, user scenario information, real-time location information, social relationships, and message sending records. Among them, the basic information of users includes name, age, gender, contact information, education, registration date, etc.; user behavior data includes user clicks, browsing, searching, purchasing, comments and ratings, user preferences and preferences, and dynamic interaction behaviors with applications. Scenario information refers to information about users in a specific environment or situation, which can include device type, network status, time period, etc. Real-time location information can be expressed in latitude and longitude coordinates or address format, such as longitude and latitude values, geographic location names, postal codes, etc. Social relationship information describes the connection and relationship between a user and other users, such as user identifiers, friend lists, followers or followers lists, etc. The message sending record contains information such as the message sending time, sending channel, sending frequency, sending content and receipt status.

[0032] In order to protect user privacy and comply with relevant laws and regulations, the source data acquisition module is also responsible for managing the privacy protection and compliance of user data, formulating data storage and processing strategies according to relevant laws and regulations, and providing security protection measures for user data to ensure the security and compliance of user data.

[0033] Step S202: Analyze and divide the users to be pushed according to the source data, and divide the users to be pushed into corresponding target life cycle stages and target groups. The target life cycle stage is any one of the full life cycle stages, and the full life cycle stages include acquisition stage, activation stage, participation and use stage, conversion stage, retention and growth stage, and churn stage. The target group is at least one of the following: active users, inactive users, and churned users.

[0034] Specifically, Figure 3As shown in the figure, the user analysis and segmentation module is responsible for analyzing and segmenting the user's behavior, interests and characteristics, and dividing the users into different life cycle stages and groups. Specifically, the source data obtained by the source data acquisition module can be used to construct an undirected weighted complex network through mutual information, and the community detection algorithm based on node similarity can be used to segment the users to be pushed into the corresponding target life cycle stages and target groups, thereby understanding the user's behavior habits, purchase history and interest preferences, and providing a basis for subsequent push decisions.

[0035] Step S203: determining the message content and message template corresponding to the user to be pushed according to the target lifecycle stage and the source data.

[0036] Specifically, Figure 3 As shown, the message content and template management module sets the corresponding message content and templates according to the target life cycle stages and target groups divided by the user analysis and segmentation module. Different message content and templates are formulated for users at different life cycle stages to ensure that the pushed messages match the needs and preferences of users. For example, in the acquisition stage: send a welcome email, SMS or pop-up window to introduce the features and advantages of the product or service, provide a registration or download link, and stimulate user interest; in the activation stage, participation and use stage, conversion stage, retention and growth stage, you can combine user characteristics and business needs and other characteristics to provide personalized recommendations and promote conversion.

[0037] Step S204, selecting a target push strategy corresponding to the user to be pushed according to the target life cycle stage and the target group, the target push strategy being any push strategy including a push channel, a push frequency and a push time.

[0038] Specifically, Figure 3 As shown in Figure 1, the push strategy and device management module is responsible for managing push strategies and devices. The push strategy includes push channels (such as email, SMS, WeChat, mobile app notifications, pop-up windows, etc.), push time (such as morning, afternoon, evening, etc.) and push frequency (such as three times a day, once a day, etc.). According to the target life cycle stage and target group of the user to be pushed, select the appropriate push channel, push time and push frequency to ensure that the message can be accurately and timely delivered to the target user and arouse the user's interest.

[0039] Step S205: Push the message content to the user to be pushed according to the target push strategy and the message template.

[0040] Specifically, according to the needs and behaviors of the users at different stages, personalized and highly relevant message content is sent to the corresponding users according to the target push strategy and message template. This helps to provide a better user experience for the users, meet the needs of the users at a specific stage, increase user participation and satisfaction, improve user retention rate, and promote user return and resurrection. In addition, message push based on the user's entire life cycle is also a key means of personalized push and improving conversion rate, which helps to establish and maintain a solid relationship with users.

[0041] In this embodiment, first, the source data of the user to be pushed is obtained, and the above source data at least includes user basic information, user behavior data, user scenario information, real-time location information, social relationship and message sending record; then, the above user to be pushed is analyzed and divided according to the above source data, and the above user to be pushed is divided into corresponding target life cycle stages and target groups, the above target life cycle stage is any stage in the whole life cycle stage, and the above whole life cycle stage includes acquisition stage, activation stage, participation and use stage, conversion stage, retention and growth stage and churn stage, and the above target group is at least one of the following: active users, inactive users and churned users; then, the message content and message template corresponding to the above user to be pushed are determined according to the above target life cycle stage and the above source data; then, the target push strategy corresponding to the above user to be pushed is selected according to the above target life cycle stage and the above target group, the above target push strategy is any push strategy, and the above push strategy includes push channel, push frequency and push time; finally, the above message content is pushed to the above user to be pushed according to the above target push strategy and the above message template. This application obtains the user's source data, including the basic information of the user to be pushed, user behavior data, user scenario information, real-time location information, social relationships, and message sending records, analyzes and divides the behavior, interests, and characteristics of the user to be pushed, divides the users into different life cycle stages and groups, determines the target life cycle stage and target group of the user to be pushed, and sets the corresponding message content and message template according to the divided target life cycle stage and target group to ensure that the pushed message matches the user's needs and preferences. According to the target life cycle and target group of the user to be pushed, select the appropriate push channel, push time, and push frequency to ensure that the message can be accurately and timely delivered to the user to be pushed. This application solves the problem that the prior art does not divide and classify users throughout their life cycle, fails to provide personalized, customized, and targeted message push, and leads to poor user experience for some users.

[0042] In order to enable those skilled in the art to more clearly understand the technical solution of the present application, the implementation process of the message push method based on the entire life cycle of the user of the present application will be described in detail below in combination with specific embodiments.

[0043] In order to improve the efficiency and effect of message push, in an optional implementation manner, the above step S202 includes:

[0044] Step S2021, constructing a target undirected weighted complex network based on at least the source data of the user to be pushed based on mutual information;

[0045] Step S2022: Analyze the target undirected weighted complex network using a community detection algorithm based on node similarity according to the source data, and divide all the users to be pushed into the corresponding target life cycle stages and target groups.

[0046] In the above embodiment, mutual information measures the degree of association between two variables, including not only linear relationships but also nonlinear relationships. When constructing a complex network, the mutual information between node pairs can be calculated to connect nodes with higher mutual information to construct a network with association. Here, the complex network is represented by G = (V, E), where V = {v1, v2, ..., v n} is a node set, E = {e1, e2, ..., e w} is the edge set, the number of nodes in the network is N = |V|, the number of edges is W = |E|, and the network adjacency matrix is ​​A(G) = (a ij )N×N. For an undirected weighted network, its adjacency matrix is ​​also a symmetric matrix. If there is an edge between nodes i and j, then a ij =w(i,j), otherwise a ij = 0, where w(i, j) is the weight of the edge. The strength of a node is defined as the sum of the weights of the edges connecting the node. Where V(i) represents the set of all adjacent nodes of node i. An undirected weighted network, that is, the edge weights of the network represent the different data flows of a bidirectional communication link. Because the weights of the edges in the network are different, the probability distribution of each edge of the same node will be different. At this time, the probability of the edge of the node is related to the strength of the node and the weight on the edge. For all the edges connected to node i, the probability of edge (i, j) is defined as: p ij =w(i, j) / S(i)(1), let the mutual information from node i to node j be I(i, j), because w(i, j) = w(j, i), so it is defined as follows: The amount of information of node i is the sum of the mutual information between node i and other nodes, set to I(i): From the above definition, we know that I(i, j)+I(j, i)=0, and the sum of the information of the entire network is zero. Sort the information of all nodes from large to small, and the node with larger information is more important. Formula (2) and Formula (3) meet two conditions: 1) For node i, when its own strength S(i) is larger, it indicates that the node is more important, which is in line with intuitive understanding; 2) For node i's adjacent node j, when the edge weight w(i, j) is determined, the smaller the strength S(j) of node j is, it indicates that the information allocated to node i among all adjacent nodes of node j is relatively more, that is, node i is relatively more important. After constructing the target undirected weighted complex network, the community detection algorithm based on node similarity is used to divide the users to be pushed into different life cycle stages and customer groups. By accurately locating the target user group, the efficiency and effect of message push can be improved. Provide users with content and services that better meet their needs and preferences, thereby enhancing user experience.

[0047] In order to improve the accuracy and effect of push notification, in an optional implementation manner, the above step S2021 includes:

[0048] Step S20211, a selection step, selecting a message push data set, the message push data set is divided into a global data set and a sample data set, the global data set is a collection of all source data of all users and all life cycle stages, the sample data set is a collection of the source data corresponding to all the users to be pushed in the global data set;

[0049] Step S20212, a preprocessing step, performing a preprocessing operation on the message push data set to obtain source data features, the preprocessing operation includes a data cleaning operation and a feature selection operation, and the source data features at least include social relationship features, behavior similarity, scenario matching, and location similarity;

[0050] Step S20213, a construction step, defines the user basic information in the message push data set after the preprocessing operation as a node and defines the source data features as an edge and assigns weights, and constructs a target undirected weighted complex network through mutual information. The mutual information is used to measure the degree of association between node pairs. The node pair is any two of the nodes, and one node represents one of the users to be pushed.

[0051] In the above embodiment, for the selection of the message push data set, a global data set and a sample data set can be selected. Among them, the global data set is to select a data set covering the global user group and the entire life cycle to obtain comprehensive user behavior and interest information. The sample data set is to select a suitable number of data sets from the global data set, and the sample data set at least includes the source data of all users to be pushed. The sample data should be representative, and a certain data sample can be obtained by sampling algorithm or random selection. However, when selecting data, it should be ensured that the data is balanced, that is, to ensure that the data set is balanced in various categories and features, and to avoid certain categories or features occupying too much or too little proportion in the data. After selecting the data set, the data should be preprocessed, that is, data cleaning and feature selection should be performed. Data cleaning is to ensure the quality and accuracy of the data by removing noise and abnormal data. For large-scale data sets, dimensionality reduction technology (principal component analysis) can be used to reduce the data dimension and improve the efficiency and speed of the algorithm. Regarding feature extraction, correlation analysis can be selected for feature extraction. Correlation analysis is used to analyze the correlation between features to obtain source data features. After data preprocessing, the user's basic information is used to define nodes, such as user ID or user name, and the edges are defined according to social relationships, behavioral relationships, location relationships, etc. Since the degree of association between users is different, the present invention assigns weights to edges according to the behavioral similarity, scenario matching, and location similarity between users, and constructs an undirected weighted complex network through mutual information. By measuring the degree of association between node pairs through mutual information, a more accurate and comprehensive user relationship network is obtained. By constructing a complex network, it can be found that the social relationships, behavioral similarity, scenario matching, and location similarity between users provide a more accurate data basis for subsequent recommendation algorithms and personalized push notifications, and the degree of association between users can be better understood, thereby improving the accuracy and effect of push notifications.

[0052] In order to ensure the quality and relevance of the data, in an optional implementation manner, the above step S20212 includes:

[0053] Step S202121, performing the above data cleaning operation on the above message push data set to remove noise and abnormal data, and obtain a valid message push data set;

[0054] Step S202122, extracting features from the valid message push data set using correlation analysis, and calculating the similarity between the features using correlation coefficients or covariance to obtain the source data features.

[0055] In the above embodiment, missing values, outliers, and duplicate records in the message push data set are checked. For missing values, you can choose to fill (for example, using the mean, median, or mode) or delete. For outliers, statistical methods (such as Z-score, IQR) or visualization methods (such as box plots) can be used to identify and process. Ensure that the format of all data columns is consistent, such as date format, numerical format, etc. Convert the data to a format suitable for analysis, such as converting text dates to date types and encoding categorical data into numerical types. Use statistical methods or machine learning algorithms to identify and remove noise in the data, such as using filters or clustering algorithms. The cleaned data set is more accurate and reliable, reducing the deviation of analysis results caused by data quality problems. The specific process of feature extraction is to select features that have an impact on the target variable based on business needs and data understanding. Use correlation analysis (such as Pearson correlation coefficient, Spearman rank correlation coefficient) to evaluate the linear or nonlinear relationship between features. Use correlation coefficients or covariance to calculate the similarity between features. The correlation coefficient measures the strength and direction of the linear relationship between two variables, while the covariance measures the joint change of the two variables. That is, select basic user information, user behavior data (such as clicks, browsing and searching web page URLs, purchase amounts, etc.), contextual information and location information (such as device type, network status, longitude and latitude, time period, etc.) and social relationship and message sending records (interaction frequency, relevance), and use correlation coefficients or covariance to calculate the similarity and relationship between features, and finally derive social relationship features, behavior relationship features, and location relationship features. Feature extraction can reduce the number of features required for model training and improve the efficiency of model training. Through similarity calculation, the potential relationship between features can be discovered, which is very valuable for feature engineering and model building.

[0056] In order to ensure that the target undirected weighted complex network has strong correlation, in an optional implementation manner, the above step S20213 includes:

[0057] Step S202131, a first calculation step, calculating the mutual information between any two nodes, and constructing a current undirected weighted complex network based on the mutual information;

[0058] Step S202132, a second calculation step, calculating the characteristic path length and aggregation coefficient of the current undirected weighted complex network, wherein the characteristic path length is the average value of the path lengths of all the above node pairs in the above current undirected weighted complex network, and the above aggregation coefficient is used to characterize the overlap between two adjacent nodes;

[0059] Step S202133, repeating the steps, when the characteristic path length is less than the set path length threshold and / or the aggregation coefficient is less than the aggregation coefficient threshold, determining that the current undirected weighted complex network is unqualified, and sequentially re-execute the selection step, the preprocessing step, the construction step, the first calculation step, the second calculation step and the repetition step at least once, until the characteristic path length is greater than or equal to the set path length threshold and the aggregation coefficient is greater than or equal to the aggregation coefficient threshold;

[0060] Step S202134, when the above-mentioned characteristic path length is greater than or equal to the above-mentioned set path length threshold and the above-mentioned aggregation coefficient is greater than or equal to the above-mentioned aggregation coefficient threshold, the above-mentioned current undirected weighted complex network is determined to be qualified, and the above-mentioned current undirected weighted complex network is determined as the above-mentioned target undirected weighted complex network.

[0061] In the above embodiment, the characteristic path length is defined as the minimum number of edges connecting any two nodes in the network, and the path length of the two nodes is defined as the average value of the path length of all node pairs in the network. This is the global feature of the network. The aggregation coefficient is that if a node has k edges, the maximum number of edges that may exist between the nodes (k) connected by these k edges is k(k-1) / 2, and the fraction obtained by dividing the actual number of edges by the maximum number of edges that may exist is defined as the aggregation coefficient of this node. The average value of the aggregation coefficients of all nodes is defined as the aggregation coefficient of the network. The aggregation coefficient is a local feature of the network, which reflects the overlap between adjacent nodes. The quality of the current network is evaluated based on the comparison of the characteristic path length and the aggregation coefficient with the preset threshold. If the network is unqualified (the characteristic path length is too short or the aggregation coefficient is too low), the process of network construction and evaluation needs to be re-executed. By repeating the above steps, the network structure is continuously optimized until the network reaches a satisfactory quality standard. By setting thresholds and iterative optimization, it is ensured that the network structure is neither too dispersed (the path length is too long) nor too tight (the aggregation coefficient is too high). The network is allowed to dynamically adjust according to actual data and needs to adapt to different application scenarios. When the characteristic path length and aggregation coefficient meet the preset threshold, the network structure is considered reasonable. The network that meets the conditions is determined as the final target undirected weighted complex network. Ensure that the final network structure is both efficient and has appropriate local connectivity to meet specific application needs.

[0062] In order to accurately locate different stages and group characteristics of the user life cycle, in an optional implementation, the above step S2022 includes:

[0063] Step S20221, determining a value range of a similarity threshold according to a target similarity index, and selecting a target similarity threshold within the value range, wherein the target similarity index is at least one of the following: a Salton index, a Sorensen index, and a hub promotion index, and the similarity threshold represents the similarity between any two nodes;

[0064] Step S20222, an expansion step, adding the neighbor nodes of the current expansion node that meet the first expansion condition among all neighbor nodes of the current expansion node to the current community, the current expansion node is any node in the target undirected weighted complex network, the expansion condition is that the similarity value between the neighbor node and the current expansion node is greater than the target similarity threshold, and the neighbor nodes are other nodes connected to the node;

[0065] Step S20223, an updating step, updating the current extended node in the current community, and repeatedly executing the extending step until there is no neighbor node whose similarity value with the current extended node is greater than the target similarity threshold, and the current community is used to represent any one of the stages of the full life cycle;

[0066] Step S20224, updating the target similarity threshold, repeating the expansion step and the update step at least once in sequence until the set maximum number of repetitions is reached, and obtaining remaining nodes and multiple initial communities, wherein the remaining nodes are nodes that are not divided into any of the initial communities, and the number of the initial communities is less than or equal to the number of stages in the full life cycle stage;

[0067] Step S20225, when the number of the remaining nodes is not zero, merging the remaining nodes into the adjacent initial communities according to the similarity indexes of the remaining nodes, or, when the number of the remaining nodes is not zero and the remaining nodes are not connected to any of the nodes, forming the remaining nodes into a separate community;

[0068] Step S20226, determining the full life cycle stage according to all the initial communities or all the initial communities and the single community;

[0069] Step S20227, analyzing user information in each of the above communities and dividing the users in each of the above communities into different user groups, wherein the user information at least includes user behavior information and preference information;

[0070] Step S20228, determining the community where the user to be pushed is located as the target life cycle stage, and determining the user group where the user to be pushed is located as the target group.

[0071] In the above embodiment, after constructing an undirected weighted complex network, a community detection algorithm based on node similarity is used to divide users into different life cycles and customer groups. The specific algorithm is as follows: Select a suitable similarity index S. The importance of a node in the network depends not only on the size of the node itself, but also on the degree of dependence of the surrounding nodes on it. The higher the degree of dependence, the higher the similarity between nodes. First, select a similarity index S. Here, you can choose any one of the Salton index, Sorensen index and hub promotion index, or you can choose other indicators. The specific definition and algorithm are as follows: (1) In the network, for any node i, let Γ(i) represent the set of neighbor nodes of node i. Similarly, for any node j, let Γ(j) represent the set of neighbor nodes of node j. The common neighbors of two nodes i and j are s, then the Salton index is defined as The Sorensen indicator is defined as The hub promotion index is defined as Here k i is the degree of node i, k j is the degree of node j. Then, according to the common neighbor nodes between nodes, the similarity values ​​between all pairs of nodes on the network are calculated (if there are edges between nodes but no common neighbor nodes, their similarity value S ij =0), according to the similarity value S ij The size of the similarity threshold ε determines the upper and lower limits of the value, that is, S min <ε<S max , where the minimum similarity value S min=0, the maximum similarity value is determined according to the size of the similarity index S and the actual situation. (2) Starting from any node, all neighbor nodes with a similarity greater than the target threshold ε are added to the current community, and the neighbor nodes with a similarity greater than the target threshold ε are continued to be added to the community among the neighbors that have just joined the current community, until there are no neighbor nodes with a similarity greater than the target threshold ε, thus obtaining the first community. Continue to repeat the above process starting from the nodes that have not been searched in the network until all nodes in the network that contain similarity edges are covered (at this time, there are still a certain number of remaining nodes). By changing the value of the similarity threshold ε, different numbers of initial communities and different numbers of remaining nodes can be obtained. (3) The remaining nodes are merged into adjacent communities according to the similarity. If the similarity values ​​are the same (including the case where the similarity value is 0), they are merged into the community where the maximum degree neighbor node is located; if there is no maximum degree neighbor node, random merging is performed until all nodes in the network are classified into communities (if there are scattered nodes that are not connected to any node, such nodes are grouped separately). At this time, the community detection is completed and the algorithm ends. After the above community division, users can be divided into different life cycle stages, namely large communities, which include user acquisition stage, activation stage, participation and use stage, conversion stage, retention and growth stage, and churn stage. It can also be found that the large community also contains small communities, that is, by analyzing the user's attributes, behaviors, preferences and other information at different life cycle stages, users can also be subdivided into different groups or user segments, that is, the life cycle stage community also includes different customer groups. Such hierarchical information can accurately determine the target population for push, so as to better push content in a targeted manner. Determine the community where the above-mentioned users to be pushed are located as the above-mentioned target life cycle stage, and determine the user group where the above-mentioned users to be pushed are located as the above-mentioned target group. This method can help enterprises or organizations better understand and manage user groups, and achieve more effective user management and content push by accurately locating different stages of user life cycle and group characteristics.

[0072] In order to improve the propagation effect of message push and push content more targetedly, in an optional implementation, after the above step S205, the method further includes:

[0073] Step S301, obtaining user feedback data, the user feedback data at least including the reach rate, click rate and conversion rate of the pushed users, the pushed users are the users after the message is pushed to the to-be-pushed users;

[0074] Step S302, optimizing the message push data set, the undirected weighted complex network and the community detection algorithm according to the user feedback data;

[0075] Step S303: regularly analyzing the behavior patterns of all the users to which the push notifications have been sent through the user feedback data to adjust the target push strategy.

[0076] In the above embodiment, if Figure 3 As shown in the figure, the data analysis and optimization module is responsible for analyzing push data and user feedback, and optimizing and improving them. Through message receipts, the user's behavior data, push effect, conversion rate and other indicators are tracked, and data analysis is performed regularly to discover the patterns and trends of user behavior, and to adjust and improve the push methods and push strategies. After pushing relevant information to users at different life cycle stages, the algorithm and message push data set selection can be improved and optimized by timely collecting and utilizing user feedback data, such as user reach rate, click rate, conversion rate, etc. The reach rate, click rate and conversion rate data can be obtained by tracking and counting after the message is pushed. The reach rate refers to the proportion of messages successfully delivered to user devices, the click rate refers to the proportion of users who actually click to open after receiving the message, and the conversion rate refers to the proportion of users who actually complete the expected behavior after clicking, such as purchasing goods or registering an account. According to the reach rate, click rate and conversion rate data, the time, content, method, etc. of message push can be adjusted and optimized to improve user participation and conversion rate. At the same time, undirected weighted complex networks and community detection algorithms can be used to analyze user behavior data to find out the connections and community structures between users, thereby optimizing push strategies and improving message dissemination effects and user engagement. By analyzing user behavior patterns, user preferences and habits can be discovered, thereby adjusting target push strategies. For example, the time and method of message push can be adjusted based on the time, location, device, and other information of user clicks; the push content and targets can be adjusted based on the user's conversion behavior. This can improve the effect of message push and increase user engagement and conversion rates.

[0077] It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and that, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0078] The embodiment of the present application also provides a message push device based on the entire life cycle of a user. It should be noted that the message push device based on the entire life cycle of a user in the embodiment of the present application can be used to execute the message push method based on the entire life cycle of a user provided in the embodiment of the present application. The device is used to implement the above-mentioned embodiments and preferred implementation modes, and those that have been described will not be repeated. As used below, the term "module" can implement a combination of software and / or hardware for a predetermined function. Although the device described in the following embodiments is preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceived.

[0079] The following is an introduction to the message push device based on the user's entire life cycle provided in an embodiment of the present application.

[0080] Figure 4 1 is a structural block diagram of a message push device based on the user's entire life cycle according to an embodiment of the present application. Figure 4 As shown, the device comprises:

[0081] The first acquisition unit 10 is used to acquire source data of the user to be pushed, and the above source data at least includes user basic information, user behavior data, user context information, real-time location information, social relationship and message sending record.

[0082] Specifically, Figure 3 A message push system based on the entire life cycle of a user is proposed according to an embodiment of the present application. The message push system consists of five modules, including a source data acquisition module, a user analysis and segmentation module, a message content and template management module, a push channel and device management module, and a data analysis and optimization module. The source data acquisition module can obtain the source data of all users, including user basic information, user behavior data, user scenario information, real-time location information, social relationships, and message sending records. Among them, the basic information of users includes name, age, gender, contact information, education, registration date, etc.; user behavior data includes user clicks, browsing, searching, purchasing, comments and ratings, user preferences and preferences, and dynamic interaction behaviors with applications. Scenario information refers to information about users in a specific environment or situation, which can include device type, network status, time period, etc. Real-time location information can be expressed in latitude and longitude coordinates or address format, such as longitude and latitude values, geographic location names, postal codes, etc. Social relationship information describes the connection and relationship between a user and other users, such as user identifiers, friend lists, followers or followers lists, etc. The message sending record contains information such as the message sending time, sending channel, sending frequency, sending content and receipt status.

[0083] In order to protect user privacy and comply with relevant laws and regulations, the source data acquisition module is also responsible for managing the privacy protection and compliance of user data, formulating data storage and processing strategies according to relevant laws and regulations, and providing security protection measures for user data to ensure the security and compliance of user data.

[0084] The division unit 20 is used to analyze and divide the above-mentioned users to be pushed according to the above-mentioned source data, and divide the above-mentioned users to be pushed into corresponding target life cycle stages and target groups. The above-mentioned target life cycle stage is any stage in the whole life cycle stage. The above-mentioned whole life cycle stage includes acquisition stage, activation stage, participation and use stage, conversion stage, retention and growth stage and churn stage. The above-mentioned target group is at least one of the following: active users, inactive users and churned users.

[0085] Specifically, Figure 3 As shown in the figure, the user analysis and segmentation module is responsible for analyzing and segmenting the user's behavior, interests and characteristics, and dividing the users into different life cycle stages and groups. Specifically, the source data obtained by the source data acquisition module can be used to construct an undirected weighted complex network through mutual information, and the community detection algorithm based on node similarity can be used to segment the users to be pushed into the corresponding target life cycle stages and target groups, thereby understanding the user's behavior habits, purchase history and interest preferences, and providing a basis for subsequent push decisions.

[0086] The determination unit 30 is used to determine the message content and message template corresponding to the user to be pushed according to the target life cycle stage and the source data.

[0087] Specifically, Figure 3 As shown, the message content and template management module sets the corresponding message content and templates according to the target life cycle stages and target groups divided by the user analysis and segmentation module. Different message content and templates are formulated for users at different life cycle stages to ensure that the pushed messages match the needs and preferences of users. For example, in the acquisition stage: send a welcome email, SMS or pop-up window to introduce the features and advantages of the product or service, provide a registration or download link, and stimulate user interest; in the activation stage, participation and use stage, conversion stage, retention and growth stage, you can combine user characteristics and business needs and other characteristics to provide personalized recommendations and promote conversion.

[0088] The selection unit 40 is used to select a target push strategy corresponding to the user to be pushed according to the target life cycle stage and the target group. The target push strategy is any push strategy, and the push strategy includes a push channel, a push frequency and a push time.

[0089] Specifically, Figure 3As shown in Figure 1, the push strategy and device management module is responsible for managing push strategies and devices. The push strategy includes push channels (such as email, SMS, WeChat, mobile app notifications, pop-up windows, etc.), push time (such as morning, afternoon, evening, etc.) and push frequency (such as three times a day, once a day, etc.). According to the target life cycle stage and target group of the user to be pushed, select the appropriate push channel, push time and push frequency to ensure that the message can be accurately and timely delivered to the target user and arouse the user's interest.

[0090] The push unit 50 is used to push the message content to the user to be pushed according to the target push strategy and the message template.

[0091] Specifically, according to the needs and behaviors of the users at different stages, personalized and highly relevant message content is sent to the corresponding users according to the target push strategy and message template. This helps to provide a better user experience for the users, meet the needs of the users at a specific stage, increase user participation and satisfaction, improve user retention rate, and promote user return and resurrection. In addition, message push based on the user's entire life cycle is also a key means of personalized push and improving conversion rate, which helps to establish and maintain a solid relationship with users.

[0092] In this embodiment, by obtaining the user's source data, specifically including the basic information of the user to be pushed, user behavior data, user scenario information, real-time location information, social relationships, and message sending records, the behavior, interests, and characteristics of the user to be pushed are analyzed and divided, and the users are divided into different life cycle stages and groups, and the target life cycle stage and target group of the user to be pushed are determined. According to the divided target life cycle stage and target group, the corresponding message content and message template are set to ensure that the pushed message matches the user's needs and preferences. According to the target life cycle and target group of the user to be pushed, the appropriate push channel, push time, and push frequency are selected to ensure that the message can be accurately and timely delivered to the user to be pushed. The present application solves the problem that the prior art does not divide and classify users throughout their life cycle, fails to provide personalized, customized, and targeted message push, and leads to poor user experience for some users.

[0093] In order to improve the efficiency and effect of message push, in an optional implementation manner, the division unit includes:

[0094] A construction subunit, configured to construct a target undirected weighted complex network based on mutual information at least according to the source data of the user to be pushed;

[0095] The analysis subunit is used to analyze the target undirected weighted complex network according to the source data using a community detection algorithm of node similarity, and divide all the users to be pushed into the corresponding target life cycle stages and target groups.

[0096] In the above embodiment, mutual information measures the degree of association between two variables, including not only linear relationships but also nonlinear relationships. When constructing a complex network, the mutual information between node pairs can be calculated to connect nodes with higher mutual information to construct a network with association. Here, the complex network is represented by G = (V, E), where V = {v1, v2, ..., v n} is a node set, E = {e1, e2, ..., e w} is the edge set, the number of nodes in the network is N = |V|, the number of edges is W = |E|, and the network adjacency matrix is ​​A(G) = (a ij )N×N. For an undirected weighted network, its adjacency matrix is ​​also a symmetric matrix. If there is an edge between nodes i and j, then a ij =w(i,j), otherwise a ij = 0, where w(i, j) is the weight of the edge. The strength of a node is defined as the sum of the weights of the edges connecting the node. Where V(i) represents the set of all adjacent nodes of node i. An undirected weighted network, that is, the edge weights of the network represent the different data flows of a bidirectional communication link. Because the weights of the edges in the network are different, the probability distribution of each edge of the same node will be different. At this time, the probability of the edge of the node is related to the strength of the node and the weight on the edge. For all the edges connected to node i, the probability of edge (i, j) is defined as: p ij =w(i, j) / S(i)(1), let the mutual information from node i to node j be I(i, j), because w(i, j) = w(j, i), so it is defined as follows: The amount of information of node i is the sum of the mutual information between node i and other nodes, set to I(i): From the above definition, we know that I(i, j)+I(j, i)=0, and the sum of the information of the entire network is zero. Sort the information of all nodes from large to small, and the node with larger information is more important. Formula (2) and Formula (3) meet two conditions: 1) For node i, when its own strength S(i) is larger, it indicates that the node is more important, which is in line with intuitive understanding; 2) For node i's adjacent node j, when the edge weight w(i, j) is determined, the smaller the strength S(j) of node j is, it indicates that the information allocated to node i among all adjacent nodes of node j is relatively more, that is, node i is relatively more important. After constructing the target undirected weighted complex network, the community detection algorithm based on node similarity is used to divide the users to be pushed into different life cycle stages and customer groups. By accurately locating the target user group, the efficiency and effect of message push can be improved. Provide users with content and services that better meet their needs and preferences, thereby enhancing user experience.

[0097] In order to improve the accuracy and effect of the push, in an optional implementation manner, the above-mentioned construction subunit includes:

[0098] A selection module is used to perform the selection step and select a message push data set. The message push data set is divided into a global data set and a sample data set. The global data set is a collection of all source data of all users and all life cycle stages. The sample data set is a collection of the source data corresponding to all the users to be pushed in the global data set.

[0099] A preprocessing module, used to execute a preprocessing step, perform a preprocessing operation on the message push data set to obtain source data features, the preprocessing operation includes a data cleaning operation and a feature selection operation, and the source data features include at least social relationship features, behavior similarity, scenario matching, and location similarity;

[0100] A construction module is used to execute the construction step, define the above user basic information in the above message push data set after the above preprocessing operation as a node, and define the above source data features as edges and assign weights, and construct a target undirected weighted complex network through mutual information. The above mutual information is used to measure the degree of association between node pairs. The above node pair is any two of the above nodes, and one of the above nodes represents one of the above users to be pushed.

[0101] In the above embodiment, for the selection of the message push data set, a global data set and a sample data set can be selected. Among them, the global data set is to select a data set covering the global user group and the entire life cycle to obtain comprehensive user behavior and interest information. The sample data set is to select a suitable number of data sets from the global data set, and the sample data set at least includes the source data of all users to be pushed. The sample data should be representative, and a certain data sample can be obtained by sampling algorithm or random selection. However, when selecting data, it should be ensured that the data is balanced, that is, to ensure that the data set is balanced in various categories and features, and to avoid certain categories or features occupying too much or too little proportion in the data. After selecting the data set, the data should be preprocessed, that is, data cleaning and feature selection should be performed. Data cleaning is to ensure the quality and accuracy of the data by removing noise and abnormal data. For large-scale data sets, dimensionality reduction technology (principal component analysis) can be used to reduce the data dimension and improve the efficiency and speed of the algorithm. Regarding feature extraction, correlation analysis can be selected for feature extraction. Correlation analysis is used to analyze the correlation between features to obtain source data features. After data preprocessing, the user's basic information is used to define nodes, such as user ID or user name, and the edges are defined according to social relationships, behavioral relationships, location relationships, etc. Since the degree of association between users is different, the present invention assigns weights to edges according to the behavioral similarity, scenario matching, and location similarity between users, and constructs an undirected weighted complex network through mutual information. By measuring the degree of association between node pairs through mutual information, a more accurate and comprehensive user relationship network is obtained. By constructing a complex network, it can be found that the social relationships, behavioral similarity, scenario matching, and location similarity between users provide a more accurate data basis for subsequent recommendation algorithms and personalized push notifications, and the degree of association between users can be better understood, thereby improving the accuracy and effect of push notifications.

[0102] In order to ensure the quality and relevance of the data, in an optional implementation, the preprocessing module includes:

[0103] A data cleaning submodule is used to perform the above data cleaning operation on the above message push data set to remove noise and abnormal data and obtain a valid message push data set;

[0104] The feature extraction submodule is used to extract features from the above-mentioned valid message push data set using correlation analysis, and calculate the similarity between the above-mentioned features through correlation coefficients or covariance to obtain the above-mentioned source data features.

[0105] In the above embodiment, missing values, outliers, and duplicate records in the message push data set are checked. For missing values, you can choose to fill (for example, using the mean, median, or mode) or delete. For outliers, statistical methods (such as Z-score, IQR) or visualization methods (such as box plots) can be used to identify and process. Ensure that the format of all data columns is consistent, such as date format, numerical format, etc. Convert the data to a format suitable for analysis, such as converting text dates to date types and encoding categorical data into numerical types. Use statistical methods or machine learning algorithms to identify and remove noise in the data, such as using filters or clustering algorithms. The cleaned data set is more accurate and reliable, reducing the deviation of analysis results caused by data quality problems. The specific process of feature extraction is to select features that have an impact on the target variable based on business needs and data understanding. Use correlation analysis (such as Pearson correlation coefficient, Spearman rank correlation coefficient) to evaluate the linear or nonlinear relationship between features. Use correlation coefficients or covariance to calculate the similarity between features. The correlation coefficient measures the strength and direction of the linear relationship between two variables, while the covariance measures the joint change of the two variables. That is, select basic user information, user behavior data (such as clicks, browsing and searching web page URLs, purchase amounts, etc.), contextual information and location information (such as device type, network status, longitude and latitude, time period, etc.) and social relationship and message sending records (interaction frequency, relevance), and use correlation coefficients or covariance to calculate the similarity and relationship between features, and finally derive social relationship features, behavior relationship features, and location relationship features. Feature extraction can reduce the number of features required for model training and improve the efficiency of model training. Through similarity calculation, the potential relationship between features can be discovered, which is very valuable for feature engineering and model building.

[0106] In order to ensure that the target undirected weighted complex network has strong correlation, in an optional implementation, the above-mentioned building module includes:

[0107] A first calculation submodule, used for executing a first calculation step, calculating the above mutual information between any two nodes, and constructing a current undirected weighted complex network based on the above mutual information;

[0108] A second calculation submodule is used to perform a second calculation step, to calculate a characteristic path length and a aggregation coefficient of the current undirected weighted complex network, wherein the characteristic path length is an average value of the path lengths of all the above-mentioned node pairs in the above-mentioned current undirected weighted complex network, and the above-mentioned aggregation coefficient is used to characterize the overlap between two adjacent above-mentioned nodes;

[0109] A repeating submodule, used for executing the repeating step, when the characteristic path length is less than the set path length threshold and / or the aggregation coefficient is less than the aggregation coefficient threshold, determining that the current undirected weighted complex network is unqualified, and sequentially re-executing the selection step, the preprocessing step, the construction step, the first calculation step, the second calculation step and the repeating step at least once, until the characteristic path length is greater than or equal to the set path length threshold and the aggregation coefficient is greater than or equal to the aggregation coefficient threshold;

[0110] The determination submodule is used to determine that the current undirected weighted complex network is qualified and determine the current undirected weighted complex network as the target undirected weighted complex network when the characteristic path length is greater than or equal to the set path length threshold and the aggregation coefficient is greater than or equal to the aggregation coefficient threshold.

[0111] In the above embodiment, the characteristic path length is defined as the minimum number of edges connecting any two nodes in the network, and the path length of the two nodes is defined as the average value of the path length of all node pairs in the network. This is the global feature of the network. The aggregation coefficient is that if a node has k edges, the maximum number of edges that may exist between the nodes (k) connected by these k edges is k(k-1) / 2, and the fraction obtained by dividing the actual number of edges by the maximum number of edges that may exist is defined as the aggregation coefficient of this node. The average value of the aggregation coefficients of all nodes is defined as the aggregation coefficient of the network. The aggregation coefficient is a local feature of the network, which reflects the overlap between adjacent nodes. The quality of the current network is evaluated based on the comparison of the characteristic path length and the aggregation coefficient with the preset threshold. If the network is unqualified (the characteristic path length is too short or the aggregation coefficient is too low), the process of network construction and evaluation needs to be re-executed. By repeating the above steps, the network structure is continuously optimized until the network reaches a satisfactory quality standard. By setting thresholds and iterative optimization, it is ensured that the network structure is neither too dispersed (the path length is too long) nor too tight (the aggregation coefficient is too high). The network is allowed to dynamically adjust according to actual data and needs to adapt to different application scenarios. When the characteristic path length and aggregation coefficient meet the preset threshold, the network structure is considered reasonable. The network that meets the conditions is determined as the final target undirected weighted complex network. Ensure that the final network structure is both efficient and has appropriate local connectivity to meet specific application needs.

[0112] In order to accurately locate different stages and group characteristics of the user life cycle, in an optional implementation, the above-mentioned analysis subunit includes:

[0113] A first determination module is used to determine a value range of a similarity threshold according to a target similarity index, and select a target similarity threshold within the value range, wherein the target similarity index is at least one of the following: a Salton index, a Sorensen index, and a hub promotion index, and the similarity threshold represents the similarity between any two nodes;

[0114] An expansion module, used to execute an expansion step, adding the neighbor nodes of the current expansion node that meet the first expansion condition among all neighbor nodes of the current expansion node to the current community, the current expansion node is any node in the target undirected weighted complex network, the expansion condition is that the similarity value between the neighbor node and the current expansion node is greater than the target similarity threshold, and the neighbor nodes are other nodes connected to the node;

[0115] An updating module, configured to execute an updating step, update the current extended node in the current community, and repeatedly execute the extending step until there is no neighbor node whose similarity value with the current extended node is greater than the target similarity threshold, and the current community is used to represent any one of the stages of the full life cycle;

[0116] An update repetition module is used to update the target similarity threshold, and repeat the expansion step and the update step at least once in sequence until the set maximum number of repetitions is reached, to obtain remaining nodes and multiple initial communities, wherein the remaining nodes are nodes that are not divided into any of the initial communities, and the number of the initial communities is less than or equal to the number of stages in the full life cycle stage;

[0117] a merging module, configured to, when the number of the remaining nodes is not zero, merge the remaining nodes into the adjacent initial communities according to the similarity index of the remaining nodes, or, when the number of the remaining nodes is not zero and the remaining nodes are not connected to any of the nodes, form the remaining nodes into a separate community;

[0118] A second determination module is used to determine the full life cycle stage according to all the initial communities or all the initial communities and the individual communities;

[0119] A division module, used for analyzing user information in each of the above communities and dividing the users in each of the above communities into different user groups, wherein the user information at least includes user behavior information and preference information;

[0120] The third determining module is used to determine the community where the user to be pushed is located as the target life cycle stage, and determine the user group where the user to be pushed is located as the target group.

[0121] In the above embodiment, after constructing an undirected weighted complex network, a community detection algorithm based on node similarity is used to divide users into different life cycles and customer groups. The specific algorithm is as follows: Select a suitable similarity index S. The importance of a node in the network depends not only on the size of the node itself, but also on the degree of dependence of the surrounding nodes on it. The higher the degree of dependence, the higher the similarity between nodes. First, select a similarity index S. Here, you can choose any one of the Salton index, Sorensen index and hub promotion index, or you can choose other indicators. The specific definition and algorithm are as follows: (1) In the network, for any node i, let Γ(i) represent the set of neighbor nodes of node i. Similarly, for any node j, let Γ(j) represent the set of neighbor nodes of node j. The common neighbors of two nodes i and j are s, then the Salton index is defined as The Sorensen indicator is defined as The hub promotion index is defined as Here k i is the degree of node i, k j is the degree of node j. Then, according to the common neighbor nodes between nodes, the similarity values ​​between all pairs of nodes on the network are calculated (if there are edges between nodes but no common neighbor nodes, their similarity value S ij =0), according to the similarity value S ij The size of the similarity threshold ε determines the upper and lower limits of the value, that is, S min <ε<S max , where the minimum similarity value S min=0, the maximum similarity value is determined according to the size of the similarity index S and the actual situation. (2) Starting from any node, all neighbor nodes with a similarity greater than the target threshold ε are added to the current community, and the neighbor nodes with a similarity greater than the target threshold ε are continued to be added to the community among the neighbors that have just joined the current community, until there are no neighbor nodes with a similarity greater than the target threshold ε, thus obtaining the first community. Continue to repeat the above process starting from the nodes that have not been searched in the network until all nodes in the network that contain similarity edges are covered (at this time, there are still a certain number of remaining nodes). By changing the value of the similarity threshold ε, different numbers of initial communities and different numbers of remaining nodes can be obtained. (3) The remaining nodes are merged into adjacent communities according to the similarity. If the similarity values ​​are the same (including the case where the similarity value is 0), they are merged into the community where the maximum degree neighbor node is located; if there is no maximum degree neighbor node, random merging is performed until all nodes in the network are classified into communities (if there are scattered nodes that are not connected to any node, such nodes are grouped separately). At this time, the community detection is completed and the algorithm ends. After the above community division, users can be divided into different life cycle stages, namely large communities, which include user acquisition stage, activation stage, participation and use stage, conversion stage, retention and growth stage, and churn stage. It can also be found that the large community also contains small communities, that is, by analyzing the user's attributes, behaviors, preferences and other information at different life cycle stages, users can also be subdivided into different groups or user segments, that is, the life cycle stage community also includes different customer groups. Such hierarchical information can accurately determine the target population for push, so as to better push content in a targeted manner. Determine the community where the above-mentioned users to be pushed are located as the above-mentioned target life cycle stage, and determine the user group where the above-mentioned users to be pushed are located as the above-mentioned target group. This method can help enterprises or organizations better understand and manage user groups, and achieve more effective user management and content push by accurately locating different stages of user life cycle and group characteristics.

[0122] In order to improve the propagation effect of message push and push content more targetedly, in an optional implementation manner, the device further includes:

[0123] A second acquisition unit is used to acquire user feedback data after pushing the message content to the user to be pushed according to the target push strategy and the message template, wherein the user feedback data at least includes a reach rate, a click rate and a conversion rate of the pushed user, and the pushed user is a user after the message is pushed to the user to be pushed;

[0124] An optimization unit, used to optimize the message push data set, the undirected weighted complex network and the community detection algorithm according to the above user feedback data;

[0125] The adjustment unit is used to regularly analyze the behavior patterns of all the above-mentioned pushed users through the above-mentioned user feedback data to adjust the above-mentioned target push strategy.

[0126] In the above embodiment, if Figure 3 As shown in the figure, the data analysis and optimization module is responsible for analyzing push data and user feedback, and optimizing and improving them. Through message receipts, the user's behavior data, push effect, conversion rate and other indicators are tracked, and data analysis is performed regularly to discover the patterns and trends of user behavior, and to adjust and improve the push methods and push strategies. After pushing relevant information to users at different life cycle stages, the algorithm and message push data set selection can be improved and optimized by timely collecting and utilizing user feedback data, such as user reach rate, click rate, conversion rate, etc. The reach rate, click rate and conversion rate data can be obtained by tracking and counting after the message is pushed. The reach rate refers to the proportion of messages successfully delivered to user devices, the click rate refers to the proportion of users who actually click to open after receiving the message, and the conversion rate refers to the proportion of users who actually complete the expected behavior after clicking, such as purchasing goods or registering an account. According to the reach rate, click rate and conversion rate data, the time, content, method, etc. of message push can be adjusted and optimized to improve user participation and conversion rate. At the same time, undirected weighted complex networks and community detection algorithms can be used to analyze user behavior data to find out the connections and community structures between users, thereby optimizing push strategies and improving message dissemination effects and user engagement. By analyzing user behavior patterns, user preferences and habits can be discovered, thereby adjusting target push strategies. For example, the time and method of message push can be adjusted based on the time, location, device, and other information of user clicks; the push content and targets can be adjusted based on the user's conversion behavior. This can improve the effect of message push and increase user engagement and conversion rates.

[0127] The above-mentioned message push device based on the user's entire life cycle includes a processor and a memory. The above-mentioned first acquisition unit, division unit and determination unit are all stored in the memory as program units, and the processor executes the above-mentioned program units stored in the memory to implement corresponding functions. The above-mentioned modules are all located in the same processor; or, the above-mentioned modules are located in different processors in any combination.

[0128] The processor contains a kernel, which calls the corresponding program unit from the memory. One or more kernels can be set, and the kernel parameters can be adjusted to solve the problem that the existing technology does not divide and classify users throughout their life cycle, fails to provide personalized, customized and targeted message push, and leads to poor user experience for some users.

[0129] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM, and the memory includes at least one memory chip.

[0130] An embodiment of the present invention provides a computer-readable storage medium, which includes a stored program, wherein when the program is running, the device where the computer-readable storage medium is located is controlled to execute the message push method based on the user's entire life cycle.

[0131] An embodiment of the present invention provides a processor, which is used to run a program, wherein the message push method based on the entire life cycle of a user is executed when the program is running.

[0132] An embodiment of the present invention provides a message push system, which includes a processor, a memory, and a program stored in the memory and executable on the processor. When the processor executes the program, at least the following steps are implemented:

[0133] Step S201, obtaining source data of the user to be pushed, the source data at least including basic user information, user behavior data, user context information, real-time location information, social relationships and message sending records;

[0134] Step S202, analyzing and dividing the users to be pushed according to the source data, dividing the users to be pushed into corresponding target life cycle stages and target groups, the target life cycle stage is any stage in the whole life cycle stage, the whole life cycle stage includes acquisition stage, activation stage, participation and use stage, conversion stage, retention and growth stage and churn stage, the target group is at least one of the following: active users, inactive users and churned users;

[0135] Step S203, determining the message content and message template corresponding to the user to be pushed according to the target life cycle stage and the source data;

[0136] Step S204, selecting a target push strategy corresponding to the user to be pushed according to the target life cycle stage and the target group, the target push strategy being any push strategy including a push channel, a push frequency and a push time;

[0137] Step S205: Push the message content to the user to be pushed according to the target push strategy and the message template.

[0138] The present application also provides a computer program product, which, when executed on a data processing device, is suitable for executing a program for initializing at least the following method steps:

[0139] Step S201, obtaining source data of the user to be pushed, the source data at least including basic user information, user behavior data, user context information, real-time location information, social relationships and message sending records;

[0140] Step S202, analyzing and dividing the users to be pushed according to the source data, dividing the users to be pushed into corresponding target life cycle stages and target groups, the target life cycle stage is any stage in the whole life cycle stage, the whole life cycle stage includes acquisition stage, activation stage, participation and use stage, conversion stage, retention and growth stage and churn stage, the target group is at least one of the following: active users, inactive users and churned users;

[0141] Step S203, determining the message content and message template corresponding to the user to be pushed according to the target life cycle stage and the source data;

[0142] Step S204, selecting a target push strategy corresponding to the user to be pushed according to the target life cycle stage and the target group, the target push strategy being any push strategy including a push channel, a push frequency and a push time;

[0143] Step S205: Push the message content to the user to be pushed according to the target push strategy and the message template.

[0144] Obviously, those skilled in the art should understand that the above modules or steps of the present invention can be implemented by a general computing device, they can be concentrated on a single computing device, or distributed on a network composed of multiple computing devices, they can be implemented by a program code executable by a computing device, so that they can be stored in a storage device and executed by the computing device, and in some cases, the steps shown or described can be executed in a different order than here, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. Thus, the present invention is not limited to any specific combination of hardware and software.

[0145] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0146] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0147] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0148] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0149] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0150] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0151] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0152] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity 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, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0153] From the above description, it can be seen that the above embodiments of the present application achieve the following technical effects:

[0154] 1) The message push method based on the entire life cycle of the user of the present application obtains the source data of the user, specifically including the basic information of the user to be pushed, user behavior data, user scenario information, real-time location information, social relationships and message sending records, analyzes and divides the behavior, interests and characteristics of the user to be pushed, divides the users into different life cycle stages and groups, determines the target life cycle stage and target group of the user to be pushed, and sets the corresponding message content and message template according to the divided target life cycle stage and target group to ensure that the pushed message matches the user's needs and preferences. According to the target life cycle and target group of the user to be pushed, select the appropriate push channel, push time and push frequency to ensure that the message can be accurately and timely delivered to the user to be pushed. The present application solves the problem that the prior art does not divide and classify users throughout their life cycle, fails to provide personalized, customized and targeted message push, and leads to poor user experience for some users.

[0155] 2) The message push device based on the user's entire life cycle of the present application obtains the user's source data, specifically including the basic information of the user to be pushed, user behavior data, user scenario information, real-time location information, social relationships, and message sending records, analyzes and divides the behavior, interests, and characteristics of the user to be pushed, divides the users into different life cycle stages and groups, determines the target life cycle stage and target group of the user to be pushed, and sets the corresponding message content and message template according to the divided target life cycle stage and target group to ensure that the pushed message matches the user's needs and preferences. According to the target life cycle and target group of the user to be pushed, select the appropriate push channel, push time, and push frequency to ensure that the message can be accurately and timely delivered to the user to be pushed. The present application solves the problem that the prior art does not divide and classify users throughout their life cycle, fails to provide personalized, customized, and targeted message push, and leads to poor user experience for some users.

[0156] The above description is only the preferred embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A message push method based on the user's full life cycle, characterized in that: include: Obtain source data of the user to be pushed, the source data including at least basic user information, user behavior data, user context information, real-time location information, social relationships, and message sending records; Analyze and divide the users to be pushed according to the source data, and divide the users to be pushed into corresponding target life cycle stages and target groups, wherein the target life cycle stage is any stage in the whole life cycle stage, and the whole life cycle stage includes the acquisition stage, the activation stage, the participation and use stage, the conversion stage, the retention and growth stage, and the churn stage, and the target group is at least one of the following: active users, inactive users, and churned users; Determine the message content and message template corresponding to the user to be pushed according to the target life cycle stage and the source data; Selecting a target push strategy corresponding to the user to be pushed according to the target life cycle stage and the target group, wherein the target push strategy is any push strategy, and the push strategy includes a push channel, a push frequency, and a push time; The message content is pushed to the user to be pushed according to the target push strategy and the message template.

2. The method according to claim 1, characterized in that Analyzing and dividing the users to be pushed according to the source data, and dividing the users to be pushed into corresponding target life cycle stages and target groups, including: Constructing a target undirected weighted complex network based on at least the source data of the user to be pushed based on mutual information; The target undirected weighted complex network is analyzed according to the source data using a community detection algorithm of node similarity, and all the users to be pushed are divided into the corresponding target life cycle stages and target groups.

3. The method according to claim 2, characterized in that Constructing a target undirected authorized network based on at least the source data of the user to be pushed based on mutual information, including: A selection step, selecting a message push data set, wherein the message push data set is divided into a global data set and a sample data set, wherein the global data set is a collection of all source data of all users and all life cycle stages, and the sample data set is a collection of the source data corresponding to all the users to be pushed in the global data set; A preprocessing step, performing a preprocessing operation on the message push data set to obtain source data features, wherein the preprocessing operation includes a data cleaning operation and a feature selection operation, and the source data features include at least social relationship features, behavior similarity, scenario matching, and location similarity; The construction step defines the user basic information in the message push data set after the preprocessing operation as a node and defines the source data features as an edge and assigns a weight, and constructs a target undirected weighted complex network through mutual information. The mutual information is used to measure the degree of association between node pairs. The node pair is any two nodes, and one node represents one user to be pushed.

4. The method according to claim 3, characterized in that The message push data set is preprocessed to obtain source data features, including: Perform the data cleaning operation on the message push data set to remove noise and abnormal data, and obtain a valid message push data set; The feature extraction is performed on the effective message push data set by using correlation analysis, and the similarity between the features is calculated by using correlation coefficient or covariance to obtain the source data feature.

5. The method according to claim 3, characterized in that: The target undirected weighted complex network is constructed through mutual information, including: The first calculation step is to calculate the mutual information between any two nodes, and construct a current undirected weighted complex network based on the mutual information; The second calculation step is to calculate the characteristic path length and aggregation coefficient of the current undirected weighted complex network, wherein the characteristic path length is the average value of the path lengths of all the node pairs in the current undirected weighted complex network, and the aggregation coefficient is used to characterize the overlap between two adjacent nodes; Repeating step, when the characteristic path length is less than the set path length threshold and / or the aggregation coefficient is less than the aggregation coefficient threshold, determining that the current undirected weighted complex network is unqualified, and sequentially re-execute the selection step, the preprocessing step, the construction step, the first calculation step, the second calculation step and the repeating step at least once until the characteristic path length is greater than or equal to the set path length threshold and the aggregation coefficient is greater than or equal to the aggregation coefficient threshold; When the characteristic path length is greater than or equal to the set path length threshold and the aggregation coefficient is greater than or equal to the aggregation coefficient threshold, the current undirected weighted complex network is determined to be qualified, and the current undirected weighted complex network is determined as the target undirected weighted complex network.

6. The method according to claim 2, characterized in that The target undirected weighted complex network is analyzed using a community detection algorithm of node similarity according to the source data, and all the users to be pushed are divided into the corresponding target life cycle stages and target groups, including: Determine a value range of a similarity threshold according to a target similarity index, and select a target similarity threshold within the value range, wherein the target similarity index is at least one of the following: a Salton index, a Sorensen index, and a hub promotion index, and the similarity threshold represents the similarity between any two nodes; An expansion step, adding the neighbor nodes of the current expansion node that meet the first expansion condition among all neighbor nodes of the current expansion node to the current community, the current expansion node is any node in the target undirected weighted complex network, the expansion condition is the condition that the similarity value between the neighbor node and the current expansion node is greater than the target similarity threshold, and the neighbor nodes are other nodes connected to the node; an updating step, updating the current extended node in the current community, and repeatedly executing the extending step until there is no neighbor node whose similarity value with the current extended node is greater than the target similarity threshold, and the current community is used to characterize any stage of the full life cycle stage; The target similarity threshold is updated, and the expansion step and the update step are sequentially repeated at least once until the set maximum number of repetitions is reached, to obtain remaining nodes and multiple initial communities, wherein the remaining nodes are nodes that are not divided into any of the initial communities, and the number of the initial communities is less than or equal to the number of stages in the full life cycle stage; When the number of the remaining nodes is not zero, merging the remaining nodes into adjacent initial communities according to the similarity indexes of the remaining nodes, or, when the number of the remaining nodes is not zero and the remaining nodes are not connected to any of the nodes, forming the remaining nodes into a separate community; Determining the full life cycle stage based on all the initial communities or all the initial communities and the separate community; Analyzing user information in each of the communities and dividing the users in each of the communities into different user groups, wherein the user information at least includes user behavior information and preference information; The community where the user to be pushed is located is determined as the target life cycle stage, and the user group where the user to be pushed is located is determined as the target group.

7. The method according to claim 1, characterized in that After pushing the message content to the to-be-pushed user according to the target push strategy and the message template, the method further includes: Obtaining user feedback data, the user feedback data at least including the reach rate, click rate and conversion rate of the pushed users, the pushed users being the users after the message is pushed to the to-be-pushed users; Optimizing a message push data set, an undirected weighted complex network, and a community detection algorithm according to the user feedback data; The behavior patterns of all the pushed users are regularly analyzed through the user feedback data to adjust the target push strategy.

8. A message push device based on the user's full life cycle, characterized in that: The device comprises: A first acquisition unit is used to acquire source data of the user to be pushed, wherein the source data includes at least basic user information, user behavior data, user context information, real-time location information, social relationships, and message sending records; a division unit, configured to analyze and divide the users to be pushed according to the source data, and divide the users to be pushed into corresponding target life cycle stages and target groups, wherein the target life cycle stage is any one of the full life cycle stages, and the full life cycle stages include an acquisition stage, an activation stage, a participation and use stage, a conversion stage, a retention and growth stage, and a churn stage, and the target group is at least one of the following: active users, inactive users, retained users, and churned users; A determination unit, configured to determine the message content and message template corresponding to the user to be pushed according to the target life cycle stage and the source data; A selection unit, configured to select a target push strategy corresponding to the user to be pushed according to the target life cycle stage and the target group, wherein the target push strategy is any push strategy, and the push strategy includes a push channel, a push frequency, and a push time; A push unit is used to push the message content to the user to be pushed according to the target push strategy and the message template.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored program, wherein when the program is executed, the device where the computer-readable storage medium is located is controlled to execute the method according to any one of claims 1 to 7.

10. A computer program product comprising computer instructions, characterized in that: When the computer instructions are executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images