Information push method, device, electronic device and computer program product

Abstract

The present application relates to the field of information and communication technology, and provides an information push method, device, electronic device and computer program product, the method comprising: calculating the preferred time, preferred channel information and preferred product information of each user through an information anti-disturbance model and a stream processing method; generating the anti-disturbance frequency and push strategy of each user according to the preferred channel information and preferred product information of each user; and pushing marketing information to each user according to the push strategy, preferred time and anti-disturbance frequency of each user. The information push method provided in the embodiment of the present application realizes one push strategy for one user, and at the same time combines the preferred time and anti-disturbance frequency of each user for double restrictions, thereby improving the utilization rate of information resources.

Description

Technical Field

[0001] The present application relates to the field of information and communication technology, and in particular to an information push method, device, electronic device and computer program product. Background Art

[0002] As one of the commonly used means of product promotion in the communications industry, SMS group messaging is widely used in product marketing. However, the current SMS group messaging lacks a filtering mechanism and cannot filter users who have been sent SMS group messages multiple times. That is, one push strategy corresponds to multiple users, and it is impossible to achieve one push strategy for one user. When marketing information needs to be pushed to a certain user, the marketing information is pushed to all users through SMS group messaging. Even users who are not recipients of marketing information will be forced to receive marketing information. Furthermore, the current SMS group messaging lacks management and control and can only be simply controlled based on time periods. That is, within a certain time period, marketing information can be pushed to users as many times as desired, resulting in low utilization of information resources. Summary of the invention

[0003] The present application provides an information push method, device, electronic device and computer program product, aiming to improve the utilization rate of information resources.

[0004] In a first aspect, the present application provides an information push method, comprising:

[0005] The preferred time, preferred channel information and preferred product information of each user are calculated through the information anti-disturbance model and stream processing method;

[0006] Generate the do not disturb frequency and push strategy for each user according to the preferred channel information and preferred product information of each user;

[0007] Marketing information is pushed to each of the users according to the push strategy, preferred time and anti-disturbance frequency of each of the users.

[0008] In one embodiment, the step of pushing marketing information to each user according to the push strategy, preferred time and anti-disturbance frequency of each user includes:

[0009] Determining whether the push strategy of each user meets the push requirement according to the execution period of the push strategy of each user and the anti-disturbance frequency of each user;

[0010] If it is determined that the push strategy of each of the users meets the push requirement, then the remaining monthly quota of the push strategy of each of the users is determined;

[0011] Marketing information is pushed to each of the users according to the remaining monthly quota of the push strategy of each of the users and the preferred time of each of the users.

[0012] The step of determining the remaining monthly quota of the push strategy of each user comprises:

[0013] Calculate the remaining monthly quota of batch push of the push strategy of each user according to the total number of daily contacts, the proportion of mass contact, the first preset number of days information and the mass contact upward ratio of the push strategy of each user;

[0014] According to the total daily contacts, event contact ratio, second preset days information and event contact upward ratio of the push strategy of each of the users, the remaining monthly quota of real-time push of the push strategy of each of the users is calculated.

[0015] The step of pushing marketing information to each of the users according to the remaining monthly quota of the push strategy of each of the users and the preferred time of each of the users comprises:

[0016] If it is determined that the remaining monthly quota of batch push for each of the users is greater than the total number of batch push contacts corresponding to it, and it is determined that the remaining monthly quota of real-time push for each of the users is greater than the total number of real-time push contacts corresponding to it, then determining whether the push time of the push strategy of each of the users is within the preferred time of each of the users;

[0017] If it is determined that the push time of the push strategy of each of the users is within the preferred time of each of the users, marketing information is pushed to each of the users according to the priority of the push strategy of each of the users.

[0018] The step of pushing marketing information to each of the users according to the priority of the push strategy of each of the users comprises:

[0019] Sorting the push strategies of each user from high to low according to the original priority to obtain the push strategies after the first sorting;

[0020] The push strategies after the first sorting are sorted again from high to low according to the priority set by the administrator to obtain the push strategies after the second sorting;

[0021] According to the order of the push strategies after the second sorting, marketing information is pushed to each of the users in turn.

[0022] Before the step of calculating the preferred time, preferred channel information and preferred product information of each user by using the information anti-disturbance model and the stream processing method, the method further includes:

[0023] Samples of the preset labeled data set are selected by up-sampling and down-sampling methods to obtain the data set to be processed;

[0024] Performing feature preprocessing selection on the data set to be processed by using a numerical missing value method and a category reduction method to obtain various feature preprocessing data;

[0025] Performing feature selection on each of the feature preprocessed data by using a mutual information feature selection method and a regularization feature selection method to obtain a training data set;

[0026] The preset model file is trained by using the training data set, and an initial model is obtained by combining a category weighting algorithm and a model fusion method;

[0027] The initial model is optimized according to the algorithm hyperparameters and key hyperparameters of the category weighted algorithm to obtain an information anti-disturbance model.

[0028] After the step of pushing marketing information to each of the users according to the push strategy, preferred time and anti-disturbance frequency of each of the users, the method further includes:

[0029] Recording node data of each node in the process of pushing the marketing information, wherein the node data includes the number of target users, the number of special list users, the number of quota users, the number of frequency users, the number of product users and the number of channel users;

[0030] A funnel chart is created based on the target number of users, the special list number of users, the quota number of users, the frequency number of users, the product number of users and the channel number of users, so that business personnel can determine the number of filtered users for each node based on the funnel chart.

[0031] In a second aspect, the present application also provides an information push device, including:

[0032] The inference module is used to infer the preferred time, preferred channel information and preferred product information of each user through the information anti-disturbance model and stream processing method;

[0033] A generating module, used for generating the do not disturb frequency and push strategy for each user according to the preferred channel information and preferred product information of each user;

[0034] The push module is used to push marketing information to each of the users according to the push strategy, preferred time and anti-disturbance frequency of each of the users.

[0035] In a third aspect, the present application also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the steps of the information push method described in the first aspect are implemented.

[0036] In a fourth aspect, the present application further provides a computer program product, wherein the computer program product comprises a computer program, and when the computer program is executed by the processor, the steps of the information push method described in the first aspect are implemented.

[0037] The information push method, device, electronic device and computer program product provided by the present application generate push strategies for each user according to the preferred channel information and preferred product information of each user during the information push process, thereby realizing one push strategy for each user, and eliminating the need to push marketing information to all users through the same push strategy, thereby realizing accurate push of marketing information and improving the utilization rate of information resources. At the same time, dual restrictions are imposed by combining the preferred time of each user and the anti-disturbance frequency, so that marketing information cannot be pushed to the user any number of times within a certain time period, thereby improving the utilization rate of information resources. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the technical solutions in the present application or the prior art, a brief introduction will be given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0039] Figure 1 This is one of the flow charts of the information push method provided by this application;

[0040] Figure 2 This is the second flow chart of the information push method provided by this application;

[0041] Figure 3 This is the third flow chart of the information push method provided by this application;

[0042] Figure 4 It is a structural schematic diagram of the information push device provided by this application;

[0043] Figure 5 It is a structural schematic diagram of the electronic device provided by this application. DETAILED DESCRIPTION

[0044] In order to make the purpose, technical solutions and advantages of this application clearer, the technical solutions in this application will be clearly and completely described below in conjunction with the drawings in this application. Obviously, the described embodiments are part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0045] Combine the following Figures 1 to 5The present invention describes the information push method, device, electronic device and computer program product provided by the present application.

[0046] Specifically, the present application provides an information push method, referring to Figure 1 , Figure 1 This is one of the flow charts of the information push method provided by this application.

[0047] The embodiments of the present application provide embodiments of the information push method. It should be noted that although the logical order is shown in the flow chart, under certain data, the steps shown or described may be completed in a different order than here.

[0048] The embodiments of the present application take an electronic device as an example of an execution subject. The embodiments of the present application take an information push system as one of the manifestations of the electronic device, but do not limit the electronic device.

[0049] The information push method provided in the embodiment of the present application includes:

[0050] Step S10, calculating each user's preferred time, preferred channel information and preferred product information through an information anti-disturbance model and a stream processing method.

[0051] It should be noted that before calculating each user's preferred time, preferred channel information and preferred product information, the information push system needs to build an intelligent real-time information anti-disturbance control system. The technical level of the information anti-disturbance control system can be divided into three parts: real-time access to customer contact data through stream processing methods, real-time calculation of anti-disturbance frequency and resource flow control, and construction of an information anti-disturbance model.

[0052] The specific details of real-time access to customer contact data are as follows: when contact data is generated, data is collected through kafka (open source stream processing platform). After data collection is completed, the collected data is batch processed, cleaned, calculated and loaded through the scheduling tool, and then a real-time stream computing framework is built through Spark+Flink (distributed stream data flow engine) to complete data calculation. Then, the customer contact data is received and processed through kafka, and the customer contact data is converted at the same time for offline analysis or real-time monitoring, real-time prediction, etc. The information push system calculates it once every preset time. Furthermore, each customer's real-time contact data uses a Flink task, even if it is implemented by FlinkSQL, such as consuming a Kafka Topic, calculating the customer's latest contact, ordering, complaint and other data, etc., and connecting a set of ClickHouse through Flink to realize the overall OLAP. Based on Spark+Flink, we build a data stream and batch integration platform for customers, innovate new visual data orchestration capabilities, shield data developers from data formats (such as batch, streaming; files, tables, messages, etc.), shield data processing engines (such as M / R, Spark, Flink), and shield technical operators from data developers, provide a unified orchestration experience based on the business perspective, and improve data processing efficiency.

[0053] The specific details of the real-time calculation of anti-disturbance frequency and resource flow control are as follows: AI (Artificial Intelligence) capabilities are introduced into the data center to support the construction of models such as anti-disturbance frequency prediction and optimal recommendation time identification. The model results are directly referenced by the frequency control in IOP marketing through FI, and are used for marketing activity frequency control and real-time resource flow control. When the input parameters remain unchanged, the model is provided with the ability to automatically retrain and update according to the marketing effect, and the effect data is directly used in production operations. After the user actively or passively contacts, the contact frequency and channel available resources are updated in real time according to the marketing contact data. Taking channel flow control as an example, after the user contacts the 10086 WeChat official account to apply for a 5-yuan traffic package, the channel contact data and channel resource usage are updated in real time. At the next contact time of the user, additional discounts are pushed based on the user’s latest product situation, or marketing discounts are recommended based on user preference data.

[0054] The information anti-disturbance model is constructed as described in step A to step E. The specific description of step A to step E is as follows:

[0055] Step A, selecting samples from a preset labeled data set by upsampling and downsampling methods to obtain a data set to be processed;

[0056] Step B, performing feature preprocessing and selection on the data set to be processed by using a numerical missing value method and a category reduction method to obtain various feature preprocessing data;

[0057] Step C, performing feature selection on each of the feature preprocessing data by using a mutual information feature selection method and a regularization feature selection method to obtain a training data set;

[0058] Step D, training the preset model file by using the training data set, and combining the category weighting algorithm and the model fusion method to obtain an initial model;

[0059] Step E: optimizing the initial model according to the algorithm hyperparameters and key hyperparameters of the category weighted algorithm to obtain an information anti-disturbance model.

[0060] Specifically, the information push system selects samples of the preset annotated data set by the up-down sampling method to obtain the data set to be processed. The preset annotated data set is selected in advance. The up-down sampling method in this embodiment can effectively improve the problem of reduced classification effect of imbalanced data transformation. Then, the information push system performs feature preprocessing selection on the data set to be processed by the numerical missing value method and the category reduction method to obtain each feature preprocessing number. The numerical missing value method and the category reduction method in this embodiment effectively improve the effect and performance of big data analysis. Next, the information push system performs feature selection on each feature preprocessing data by the mutual information feature selection method and the regularized feature selection method to obtain a training data set. The mutual information feature selection method and the regularized feature selection method in this embodiment can effectively alleviate the problems of business experience, simple data analysis and repeated attempts since feature selection for a long time. Then, the information push system trains the preset model file by the training data set, and combines the category weighting algorithm and the model fusion method to obtain the initial model, wherein the preset model file is pre-set. This embodiment improves the prediction effect by the model fusion method and the diversity of the category weighting algorithm. Finally, the information push system determines the algorithm hyperparameters of the category weighted algorithm based on network search, and identifies the key hyperparameters of the category weighted algorithm and the effective range of the key hyperparameters, and then optimizes the initial model through the algorithm hyperparameters, key hyperparameters and the effective range of the key hyperparameters to obtain the information anti-disturbance model. The algorithm hyperparameters and key hyperparameters in this embodiment can effectively solve the problem of disordered and repeated manual adjustment, as well as the problem of concavity and convexity.

[0061] Furthermore, the information anti-disturbance model includes a real-time data model and an offline data model. The real-time data model can be divided into three categories according to the application scenario, namely, a dynamic model, an event model and a spatiotemporal model; the offline data model can be divided into two categories according to the application scenario, namely, a user contact preference time model and a user service subscription preference cycle model.

[0062] Furthermore, the dynamic model is used for real-time summary statistics and supports offline operations of real-time data, such as report indicators. The event model is used to abstract real-time customer contact data into a series of events, and to mine marketing and real-time push services. The spatiotemporal model is used for channel residence, and the user's channel preference information is calculated based on the user's channel residence time.

[0063] Furthermore, the user contact preference time model is used to continuously cluster users based on the user contact behavior data on passive touchpoints (mobile APP applications, 10086 public accounts, Migu videos) through the machine modeling platform, to mine the user's access preference time, and form a regularly refreshed preference model, which is applied to the automated scenario strategy module, and combined with the user's preferred product information and preferred channel information, to provide a push strategy. An application scenario is as follows: If a user orders iQiyi's targeted traffic package between the 20th and 25th of each month for three consecutive months, similar traffic packages or other targeted benefit packages can be recommended to the user in advance within the user's preference cycle to improve user stickiness and satisfaction. The user's service subscription preference cycle model is used to predict the customer's preference for the subscription time cycle based on the user's service handling records. By using the Ranking algorithm to perform mechanical energy clustering analysis on the user's service subscription habits in different months and holidays, daily and holiday model subscription preference data are formed respectively, which are used for daily and holiday service subscription preference recommendations. An example of an application scenario: If a user logs in to the mobile app to check his or her consumption bill between the 3rd and 5th of the month for three consecutive months, the corresponding marketing products can be recommended in the mobile app during the user's monthly login period to improve the utilization rate of marketing resources and reduce the waste of marketing resources.

[0064] Furthermore, the algorithm of the information anti-disturbance model is described as follows. Taking the construction of a potential complaint user clustering model as an example, based on the user complaint contact data in April 2021, mining and analysis were carried out, and the 20% of users with the highest complaint and SMS contact coefficients through the 10086 channel were selected (including 48,000 complaining users, 24,000 contact users, and a total of 72,000 users) for clustering and grouping, and 5 groups were obtained. The group numbers of the 5 groups are group number 0, group number 1, group number 2, group number 3 and group number 4 respectively. The TOP3 feature data of the 5 groups and their feature indexes are shown in Table 1. Table 1 shows the TOP3 feature data of the 5 groups and their feature indexes.

[0065] Table 1 TOP3 characteristic data and characteristic index of the five groups

[0066]

[0067] According to the analysis of the TOP3 feature data and its feature index, the differences between user groups are mainly reflected in the dimensions of whether the user is a dual-card user that month, contact status, number of recommendations, subscription status, over-package charges, contact time, etc. The specific difference performance results are shown in Table 2. Table 2 shows the difference performance results.

[0068] Table 2 Difference performance results

[0069]

[0070] Through the machine modeling platform, we continuously conduct cluster analysis and user feature mining on users to build a potential complaint user clustering model. Similarly, the construction process of other real-time data models, user contact preference time period models, and user business subscription preference cycle models are similar.

[0071] Furthermore, after the information anti-disturbance control system is built, that is, after the information anti-disturbance model is built, the information push system determines the contact behavior data of each user, and infers the contact behavior data of each user according to the information anti-disturbance model to obtain the preferred time, preferred channel information and preferred product information of each user. Then, the information push system dynamically updates the preferred time, preferred channel information and preferred product information through a stream processing method and provides services to the outside in a real-time interface manner. When each user contacts, the latest preferred time, latest preferred channel information and latest preferred product information of each user are queried in real time.

[0072] In this embodiment, for example, the contact behavior data of user 1 is: recharging iQiyi video membership through Guangdong Mobile APP from the 1st to the 3rd of each month for three consecutive months, then the preferred time of user 1 is from the 1st to the 3rd of each month, the preferred channel information is Guangdong Mobile APP, and the preferred product information is iQiyi video membership. The contact behavior data of user 2 is: checking mobile phone bills through Guangdong Mobile official account on the 15th of each month for three consecutive months, then the preferred time of user 2 is the 15th of each month, the preferred channel information is Guangdong Mobile official account, and the preferred product information is mobile phone bill inquiry.

[0073] Step S20: generating a do not disturb frequency and a push strategy for each user according to the preferred channel information and preferred product information of each user.

[0074] The information push system analyzes the preferred channel information and preferred product information of each user through the user contact preference time model to generate the anti-disturbance frequency and push strategy for each user. It should be noted that the anti-disturbance frequency and push strategy generated in this embodiment are the latest anti-disturbance frequency and the latest push strategy obtained through the stream processing method. It can be further understood that the information push system analyzes the latest preferred channel information and the latest preferred product information of each user through the user contact preference time model to generate the latest anti-disturbance frequency and the latest push strategy for each user.

[0075] In this embodiment, for example, the preferred channel information of user 1 is 10086 public account, and the preferred product information is iQiyi video membership. Since the recharge period of video membership is once a month, the frequency of user 1's anti-disturbance is once a month, and the push strategy of user 1 is to push marketing information of iQiyi video membership recharge to user 1 through 10086 public account every other month. The preferred channel information of user 2 is Guangdong Mobile APP, and the preferred product information is 7-day traffic overlay package. Since the recharge period of 7-day traffic overlay package is 3 times a month, the frequency of user 2's anti-disturbance is 3 times a month, and the push strategy of user 2 is to push marketing information of 7-day traffic overlay package recharge to user 2 through Guangdong Mobile APP every 7 days.

[0076] Furthermore, this embodiment also supports rule contact, such as contacting a preset number of times or customer replying a preset number of times, that is, changing the push strategy, where the preset number of times is set according to the actual situation. The methods for changing the push strategy include but are not limited to the following three: The first is the display number strategy, that is, all push strategies associated with the mass-messaging channel recommend one push strategy each time according to the priority order of the strategy, and recommend N (set according to the actual situation) times to change the push strategy. The second is the completion strategy, that is, if the number of information contents in a recommended push strategy is less than the number of recommendations returned, the completion strategy is used to complete it. The third is the filtering strategy, that is, filtering the push strategy according to the status of user feedback and the number of recommendations for each marketing information.

[0077] The embodiment of the present application implements a push strategy corresponding to one user, so that users who are not recipients of marketing information will not be forced to receive marketing information, thereby reducing disturbance to users, greatly improving user satisfaction, and reducing user complaint rates.

[0078] Step S30: Push marketing information to each user according to the push strategy, preferred time and anti-disturbance frequency of each user.

[0079] When the information push system receives an execution instruction for executing the push policy of each user, it determines whether the execution period of the push policy of each user meets the anti-disturbance frequency of each user, and obtains a corresponding determination result, wherein the determination result may be that the execution period of the push policy of each user does not meet the anti-disturbance frequency of each user, or the determination result may be that the execution period of the push policy of each user meets the anti-disturbance frequency of each user. In this embodiment, for example, the anti-disturbance frequency of user 1 is once a month, and the execution period of the push policy of user 1 is determined to be the first time this month, then it is determined that the execution period of the push policy meets the anti-disturbance frequency of user 1. When the execution period of the push policy of user 1 is determined to be the second time this month, then it is determined that the execution period of the push policy does not meet the anti-disturbance frequency of user 1.

[0080] If the result of the determination is that the execution period of the push policy of each user does not meet the anti-disturbance frequency of each user, the information push system refuses to release the push policy of each user, that is, does not execute the push policy of each user. If the result of the determination is that the execution period of the push policy of each user meets the anti-disturbance frequency of each user, the information push system needs to determine the push time of the push policy of each user, and determine whether to push marketing information to each user based on the push time of the push policy of each user and the preferred time of each user, as described in steps S301 to S303.

[0081] This embodiment provides an information push method. In the process of information push, a push strategy for each user is generated according to the preferred channel information and preferred product information of each user, so that one push strategy corresponds to one user, and there is no need to push marketing information to all users through the same push strategy, so that accurate push of marketing information is achieved, and the utilization rate of information resources is improved. At the same time, dual restrictions are imposed by combining the preferred time of each user and the anti-disturbance frequency, so that marketing information cannot be pushed to the user any number of times within a certain time period, thereby improving the utilization rate of information resources.

[0082] Further, refer to Figure 2 , Figure 2 This is the second flow chart of the information push method provided by the present application, and the step S30 includes:

[0083] Step S301, determining whether the push policy of each user meets the push requirement according to the execution cycle of the push policy of each user and the anti-disturbance frequency of each user;

[0084] Step S302, if it is determined that the push policy of each user meets the push requirement, then determining the remaining monthly quota of the push policy of each user;

[0085] Step S303: Push marketing information to each of the users according to the remaining monthly quota of the push policy of each of the users and the preferred time of each of the users.

[0086] Specifically, when the information push system receives an execution instruction for executing the push strategy of each user, it determines whether the execution cycle of the push strategy of each user meets the anti-disturbance frequency of each user, that is, determines whether the push strategy of each user meets the push requirements. If it is determined that the execution cycle of the push strategy of each user meets the anti-disturbance frequency of each user, the information push system determines that the push strategy of each user meets the push requirements, and determines the regional information, channel information and the number of days in the current month of the push strategy of each user, wherein the number of days information includes but is not limited to the number of days in the current month, the number of busy days at the beginning of the month and the number of busy days at the end of the month. The number of busy days at the beginning of the month and the number of busy days at the end of the month are set according to actual conditions. By default, the default value of the number of busy days at the beginning of the month is 4 days, indicating that the first 4 days at the beginning of the month are the busy time of the month. The default value of the number of busy days at the end of the month is 1 day, indicating that the last day at the end of the month is the busy time of the month.

[0087] Next, the information push system determines the remaining monthly quota of the push strategy of each user in the corresponding region and channel for the current month according to the regional information, channel information and the number of days in the current month of the push strategy of each user, as described in step S3021 to step S3022. Finally, the information push system verifies the remaining monthly quota of the push strategy of each user for the current month to obtain a verification result, wherein the verification result may be insufficient remaining monthly quota, or sufficient remaining monthly quota. If the verification result is determined to be insufficient remaining monthly quota, the information push system determines that the remaining monthly quota is insufficient to execute the push strategy of each user. If the verification result is determined to be sufficient remaining monthly quota, the information push system determines whether the push time of the push strategy of each user is within the preferred time of each user, obtains the corresponding determination result, and determines whether to push marketing information to each user according to the determination result, as described in step S3031 to step S3032, wherein the determination result may be that the push time of the push strategy of each user is within the preferred time of each user, or the determination result may be that the push time of the push strategy of each user is not within the preferred time of each user.

[0088] In the process of pushing marketing information to each user, the embodiment of the present application combines the preferred time of each user, the frequency of interruptions, and the remaining monthly quota of the push strategy to impose multiple restrictions, so that marketing information cannot be pushed to the user arbitrarily many times, thereby improving the utilization rate of information resources.

[0089] Further, the specific description of step S3021 to step S3022 is as follows:

[0090] Step S3021, calculating the remaining monthly quota of batch push of the push strategy of each user according to the total number of daily contacts, the proportion of mass contact, the first preset number of days information and the mass contact upward ratio of the push strategy of each user;

[0091] Step S3022, calculating the remaining monthly quota of real-time push of each user's push strategy according to the total daily contact amount, event contact proportion, second preset day information and event contact upward ratio of each user's push strategy.

[0092] It should be noted that the remaining monthly quota of the push policy for the current month includes the remaining monthly quota for batch push and the remaining monthly quota for real-time push.

[0093] Specifically, the information push system obtains the total number of daily contacts, the proportion of mass contact, the first preset number of days information and the floating ratio of mass contact of each user's push strategy, wherein the first preset number of days information includes the number of days in the current month, the number of busy days at the beginning of the month and the number of busy days at the end of the month. Then, the information push system multiplies the total number of daily contacts, the proportion of mass contact, the floating ratio of mass contact, and the difference between the number of days in the current month, the number of busy days at the beginning of the month and the number of busy days at the end of the month to obtain the remaining monthly quota of batch push of each user's push strategy, and the calculation formula is as follows: Remaining monthly quota of batch push = total number of daily contacts * proportion of mass contact * (number of days in the current month - number of busy days at the beginning of the month - number of busy days at the end of the month) * (1 + floating ratio of mass contact). Further, the information push system obtains the total number of daily contacts, the proportion of event contacts, the second preset number of days information and the floating ratio of event contacts of each user's push strategy, wherein the second preset number of days information is the number of days in the current month. Next, the information push system multiplies the total daily contacts, the proportion of event contacts, the number of days in the month and the event contact floating ratio to obtain the remaining monthly quota of real-time push for each user's push strategy. The calculation formula is as follows: Remaining monthly quota of real-time push = total daily contacts * proportion of event contacts * number of days in the month * (1 + event contact floating ratio). It should be noted that when calculating the remaining monthly quota of batch push and the remaining monthly quota of real-time push, if the calculation result has decimal places, the decimal places will be discarded and the integer will be retained.

[0094] The embodiment of the present application jointly determines the remaining monthly quota of the push strategy of each user based on the remaining monthly quota of batch push and the remaining monthly quota of real-time push of the push strategy of each user, so that the remaining monthly quota of the push strategy of each user can represent the vast majority of occurrences, making the remaining monthly quota of the push strategy of each user more accurate and representative, providing a premise and guarantee for the accuracy of whether to push marketing information to each user in the future.

[0095] Further, the specific description of step S3031 to step S3032 is as follows:

[0096] Step S3031, if it is determined that the remaining monthly quota of batch push for each user is greater than the total number of batch push contacts corresponding to it, and it is determined that the remaining monthly quota of real-time push for each user is greater than the total number of real-time push contacts corresponding to it, then determine whether the push time of the push strategy of each user is within the preferred time of each user;

[0097] Step S3032: If it is determined that the push time of the push strategy of each user is within the preferred time of each user, marketing information is pushed to each user according to the priority of the push strategy of each user.

[0098] Specifically, the information push system calculates the total number of batch push contacts and the total number of real-time push contacts of the push strategy of each user, and the specific calculation formula is as follows: total number of batch push contacts = number of customer groups * cumulative number of contacts of each user in the push strategy. Total number of real-time push contacts = number of customer groups * cumulative number of contacts of each user in the strategy. Next, the information push system verifies the remaining monthly quota of the push strategy of each user according to the total number of batch push contacts and the total number of real-time push contacts of the push strategy of each user, and specifically determines whether the remaining monthly quota of batch push of the push strategy of each user is greater than the total number of batch push contacts of the push strategy of each user, and determines whether the remaining monthly quota of real-time push of the push strategy of each user is greater than the total number of real-time push contacts of the push strategy of each user. If it is determined that the remaining monthly quota of batch push of the push strategy of each user is greater than the total number of batch push contacts of the push strategy of each user, and it is determined that the remaining monthly quota of real-time push of the push strategy of each user is greater than the total number of real-time push contacts of the push strategy of each user, the information push system determines whether the push time of the push strategy of each user is within the preferred time of each user. If it is determined that the push time of each user's push strategy is within the preferred time of each user, the information push system pushes marketing information to each user according to the priority of each user's push strategy, as described in steps S30321 to S30323. If it is determined that the remaining monthly quota of batch push of each user's push strategy is less than or equal to the total number of batch push contacts of each user's push strategy, or if it is determined that the remaining monthly quota of real-time push of each user's push strategy is less than or equal to the total number of real-time push contacts of each user's push strategy, the information push system determines that the push strategy of each user does not meet the push requirements.

[0099] In the process of pushing marketing information to each user, the embodiment of the present application combines the remaining monthly quota of batch push with the total number of batch push contacts, the remaining monthly quota of real-time push with the total number of real-time push contacts, and the preferred time to impose multiple restrictions, so that marketing information cannot be pushed to the user arbitrarily many times, thereby improving the utilization rate of information resources.

[0100] Further, the specific description of step S30321 to step S30323 is as follows:

[0101] Step S30321, sorting the push strategies of each user from high to low according to the original priority, to obtain the push strategies after the first sorting;

[0102] Step S30322, re-sorting the push strategies after the first sorting from high to low according to the priority set by the administrator to obtain the push strategies after the second sorting;

[0103] Step S30323: Push marketing information to each of the users in turn according to the order of the push strategies after the second sorting.

[0104] Specifically, the information push system determines the original priority of the push strategy of each user, wherein the original priority is the priority of the marketing information creator's intention, and sorts the push strategy of each user from high to low according to the original priority to obtain the push strategy after the first sorting. Of course, this embodiment can also sort from low to high according to the original priority. Next, the information push system determines whether there is an administrator-level setting instruction at present. If it is determined that there is no administrator-level setting instruction, the information push system pushes marketing information to each user in sequence according to the order of the push strategy after the first sorting. If it is determined that there is an administrator-level setting instruction, the information push system determines the administrator-set priority of the administrator's push strategy for each user according to the administrator-level setting instruction, and sorts the push strategy after the first sorting again from high to low according to the administrator-set priority, to obtain the push strategy after the second sorting. Of course, this embodiment can also sort according to the administrator-set priority. Finally, the information push system pushes marketing information to each user in sequence according to the order of the push strategy after the second sorting.

[0105] The embodiment of the present application determines the order of push strategies through the original priority of the push strategies and the priority set by the administrator, and pushes marketing information to each user according to the order of priority, so that batch group sending is more reasonable and more orderly controlled.

[0106] Further, refer to Figure 3 , Figure 3 This is the third flow chart of the information push method provided by the present application. After step S30, it also includes:

[0107] Step S40, recording node data of each node in the process of pushing the marketing information, wherein the node data includes the number of target users, the number of special list users, the number of quota users, the number of frequency users, the number of product users and the number of channel users;

[0108] Step S50, creating a funnel chart according to the number of target users, the number of special list users, the number of quota users, the number of frequency users, the number of product users and the number of channel users;

[0109] Step S60, determining the number of filtered users of each node according to the funnel diagram.

[0110] In the process of pushing marketing information to each user, the information push system records the node data of each node in the process of pushing marketing information. The node data includes the number of target users, the number of special list users, the number of quota users, the number of frequency users, the number of product users, and the number of channel users. Then, the information push system sorts the number of target users, the number of special list users, the number of quota users, the number of frequency users, the number of product users, and the number of channel users from top to bottom or from bottom to top to create a funnel chart. Finally, the information push system determines the number of filtered users of the customer group at each node based on the funnel chart, and sends the number of filtered users at each node to the user terminal of the business personnel. Furthermore, the information push system can also send the funnel chart to the user terminal of the business personnel, and the business personnel understand the number of filtered users of the customer group at each node based on the funnel chart in the user terminal.

[0111] In this embodiment, for example, the number of target users is 342346, the number of special list users is 271734, the number of quota users is 237892, the number of frequency users is 221654, the number of product users is 81534 and the number of channel users is 17353. Then, according to the funnel chart, the number of filtered users of the special list user node is 70612, the number of filtered users of the quota user node is 33842, the number of filtered users of the frequency user node is 16238, the number of filtered users of the product user node is 140120, and the number of filtered users of the channel user node is 64181.

[0112] This embodiment provides an information push method, which realizes the visualization of the whole process of information anti-disturbance management and control through a funnel chart, clearly understands the number of users of each node and the number of filtered users of each node, so that the anti-disturbance filtering situation during the information push process can be understood in real time, and further enables the channel anti-disturbance situation during the information push process to be dynamically monitored.

[0113] Furthermore, the information push device provided in the present application is described below. The information push device described below and the information push method described above can refer to each other.

[0114] like Figure 4 As shown, Figure 4 It is a structural diagram of the information push device provided by the present application, and the information push device includes:

[0115] The calculation module 401 is used to calculate the preferred time, preferred channel information and preferred product information of each user through the information anti-disturbance model and the stream processing method;

[0116] A generating module 402, configured to generate a do not disturb frequency and a push strategy for each user according to the preferred channel information and preferred product information of each user;

[0117] The push module 403 is used to push marketing information to each of the users according to the push strategy, preferred time and anti-disturbance frequency of each of the users.

[0118] Furthermore, the push module 403 is also used for:

[0119] Determining whether the push strategy of each user meets the push requirement according to the execution cycle of the push strategy of each user and the anti-disturbance frequency of each user;

[0120] If it is determined that the push strategy of each of the users meets the push requirement, then the remaining monthly quota of the push strategy of each of the users is determined;

[0121] Marketing information is pushed to each of the users according to the remaining monthly quota of the push strategy of each of the users and the preferred time of each of the users.

[0122] Furthermore, the push module 403 is also used for:

[0123] Calculate the remaining monthly quota of batch push of the push strategy of each user according to the total number of daily contacts, the proportion of mass contact, the first preset number of days information and the mass contact upward ratio of the push strategy of each user;

[0124] According to the total daily contacts, event contact ratio, second preset days information and event contact upward ratio of the push strategy of each of the users, the remaining monthly quota of real-time push of the push strategy of each of the users is calculated.

[0125] Furthermore, the push module 403 is also used for:

[0126] If it is determined that the remaining monthly quota of batch push for each of the users is greater than the total number of batch push contacts corresponding to it, and it is determined that the remaining monthly quota of real-time push for each of the users is greater than the total number of real-time push contacts corresponding to it, then determining whether the push time of the push strategy of each of the users is within the preferred time of each of the users;

[0127] If it is determined that the push time of the push strategy of each of the users is within the preferred time of each of the users, marketing information is pushed to each of the users according to the priority of the push strategy of each of the users.

[0128] Furthermore, the push module 403 is also used for:

[0129] Sorting the push strategies of each user from high to low according to the original priority to obtain the push strategies after the first sorting;

[0130] The push strategies after the first sorting are sorted again from high to low according to the priority set by the administrator to obtain the push strategies after the second sorting;

[0131] According to the order of the push strategies after the second sorting, marketing information is pushed to each of the users in turn.

[0132] Furthermore, the information push device further includes: a model building module; the model building module is used to:

[0133] Samples of the preset labeled data set are selected by up-sampling and down-sampling methods to obtain the data set to be processed;

[0134] Performing feature preprocessing selection on the data set to be processed by using a numerical missing value method and a category reduction method to obtain various feature preprocessing data;

[0135] Performing feature selection on each of the feature preprocessed data by using a mutual information feature selection method and a regularization feature selection method to obtain a training data set;

[0136] The preset model file is trained by using the training data set, and an initial model is obtained by combining a category weighting algorithm and a model fusion method;

[0137] The initial model is optimized according to the algorithm hyperparameters and key hyperparameters of the category weighted algorithm to obtain an information anti-disturbance model.

[0138] Furthermore, the information push device further includes: a determination module; the determination module is used to:

[0139] Recording node data of each node in the process of pushing the marketing information, wherein the node data includes the number of target users, the number of special list users, the number of quota users, the number of frequency users, the number of product users and the number of channel users;

[0140] Creating a funnel chart according to the number of target users, the number of special list users, the number of quota users, the number of frequency users, the number of product users, and the number of channel users;

[0141] The number of filtered users of each of the nodes is determined according to the funnel diagram.

[0142] The specific embodiments of the information push device provided in this application are basically the same as the embodiments of the above-mentioned information push method, and will not be described in detail here.

[0143] Figure 5 An example of a physical structure diagram of an electronic device is shown in FIG. Figure 5 As shown, the electronic device may include: a processor 510, a communication interface 520, a memory 530 and a communication bus 540, wherein the processor 510, the communication interface 520 and the memory 530 communicate with each other through the communication bus 540. The processor 510 may call the logic instructions in the memory 530 to execute the information push method, which includes:

[0144] The preferred time, preferred channel information and preferred product information of each user are calculated through the information anti-disturbance model and stream processing method;

[0145] Generate the do not disturb frequency and push strategy for each user according to the preferred channel information and preferred product information of each user;

[0146] Marketing information is pushed to each user according to the push strategy, preferred time and anti-disturbance frequency of each user.

[0147] In addition, the logic instructions in the above-mentioned memory 530 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on this understanding, the technical solution of the present application can be essentially or partly embodied in the form of a software product that contributes to the prior art, and the computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a disk or an optical disk.

[0148] On the other hand, the present application further provides a computer program product, the computer program product comprising a computer program stored on a non-transitory computer-readable storage medium, the computer program comprising program instructions, and when the program instructions are executed by a computer, the computer can execute the information push method provided by the above methods, the method comprising:

[0149] The preferred time, preferred channel information and preferred product information of each user are calculated through the information anti-disturbance model and stream processing method;

[0150] Generate the do not disturb frequency and push strategy for each user according to the preferred channel information and preferred product information of each user;

[0151] Marketing information is pushed to each user according to the push strategy, preferred time and anti-disturbance frequency of each user.

[0152] In another aspect, the present application further provides a non-transitory computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the computer program is implemented to execute the above-mentioned information push method, the method comprising:

[0153] The preferred time, preferred channel information and preferred product information of each user are calculated through the information anti-disturbance model and stream processing method;

[0154] Generate the do not disturb frequency and push strategy for each user according to the preferred channel information and preferred product information of each user;

[0155] Marketing information is pushed to each user according to the push strategy, preferred time and anti-disturbance frequency of each user.

[0156] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0157] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0158] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit it. Although the present application is described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. An information push method, characterized in that: include: The preferred time, preferred channel information and preferred product information of each user are calculated through the information anti-disturbance model and stream processing method; Generate the do not disturb frequency and push strategy for each user according to the preferred channel information and preferred product information of each user; Push marketing information to each user according to the push strategy, preferred time and anti-disturbance frequency of each user; Before the step of calculating the preferred time, preferred channel information and preferred product information of each user by using the information anti-disturbance model and the stream processing method, the method further includes: Samples of the preset labeled data set are selected by up-sampling and down-sampling methods to obtain the data set to be processed; Performing feature preprocessing selection on the data set to be processed by using a numerical missing value method and a category reduction method to obtain various feature preprocessing data; Performing feature selection on each of the feature preprocessed data by using a mutual information feature selection method and a regularization feature selection method to obtain a training data set; The preset model file is trained by using the training data set, and an initial model is obtained by combining a category weighting algorithm and a model fusion method; The initial model is optimized according to the algorithm hyperparameters and key hyperparameters of the category weighted algorithm to obtain an information anti-disturbance model; Wherein, the information anti-disturbance model includes a real-time data model and an offline data model; The real-time data model includes a dynamic model, an event model and a spatiotemporal model. The dynamic model is used for real-time summary statistics, the event model is used for abstracting real-time user contact data into events, and the spatiotemporal model is used for calculating the user's channel preference information based on the user's channel residence time; The offline data model includes a user contact preference time model and a user service subscription preference cycle model. The user contact preference time model is used to perform cluster analysis on the user's contact behavior data on passive contacts to determine the user's access preference time. The user service subscription preference cycle model is used to perform cluster analysis on the user's service handling records to predict the user's preference for the subscription time cycle.

2. The information push method according to claim 1, characterized in that: The step of pushing marketing information to each user according to the push strategy, preferred time and anti-disturbance frequency of each user includes: Determining whether the push strategy of each user meets the push requirement according to the execution cycle of the push strategy of each user and the anti-disturbance frequency of each user; If it is determined that the push strategy of each of the users meets the push requirement, then determining the remaining monthly quota of the push strategy of each of the users; Marketing information is pushed to each of the users according to the remaining monthly quota of the push strategy of each of the users and the preferred time of each of the users.

3. The information push method according to claim 2, characterized in that: The step of determining the remaining monthly quota of the push strategy of each user comprises: Calculate the remaining monthly quota of batch push of the push strategy of each user according to the total number of daily contacts, the proportion of mass contact, the first preset number of days information and the mass contact upward ratio of the push strategy of each user; According to the total daily contacts, event contact ratio, second preset days information and event contact upward ratio of the push strategy of each of the users, the remaining monthly quota of real-time push of the push strategy of each of the users is calculated.

4. The information push method according to claim 2, characterized in that: The step of pushing marketing information to each of the users according to the remaining monthly quota of the push strategy of each of the users and the preferred time of each of the users comprises: If it is determined that the remaining monthly quota of batch push for each of the users is greater than the total number of batch push contacts corresponding to it, and it is determined that the remaining monthly quota of real-time push for each of the users is greater than the total number of real-time push contacts corresponding to it, then determining whether the push time of the push strategy of each of the users is within the preferred time of each of the users; If it is determined that the push time of the push strategy of each of the users is within the preferred time of each of the users, marketing information is pushed to each of the users according to the priority of the push strategy of each of the users.

5. The information push method according to claim 4, characterized in that: The step of pushing marketing information to each of the users according to the priority of the push strategy of each of the users comprises: Sorting the push strategies of each user from high to low according to the original priority to obtain the push strategies after the first sorting; The push strategies after the first sorting are sorted again from high to low according to the priority set by the administrator to obtain the push strategies after the second sorting; According to the order of the push strategies after the second sorting, marketing information is pushed to each of the users in turn.

6. The information push method according to any one of claims 1 to 5, characterized in that: After the step of pushing marketing information to each of the users according to the push strategy, preferred time and anti-disturbance frequency of each of the users, the method further includes: Recording node data of each node in the process of pushing the marketing information, wherein the node data includes the number of target users, the number of special list users, the number of quota users, the number of frequency users, the number of product users and the number of channel users; Creating a funnel chart according to the number of target users, the number of special list users, the number of quota users, the number of frequency users, the number of product users, and the number of channel users; The number of filtered users of each of the nodes is determined according to the funnel diagram.

7. An information push device, characterized in that: The information push method according to claim 1 is applied, comprising: The inference module is used to infer the preferred time, preferred channel information and preferred product information of each user through the information anti-disturbance model and stream processing method; A generating module, used for generating the do not disturb frequency and push strategy for each user according to the preferred channel information and preferred product information of each user; The push module is used to push marketing information to each of the users according to the push strategy, preferred time and anti-disturbance frequency of each of the users.

8. An electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the steps of the information push method according to any one of claims 1 to 6 are implemented.

9. A computer program product, comprising a computer program, characterized in that: When the computer program is executed by a processor, the steps of the information push method according to any one of claims 1 to 6 are implemented.

Images