Package recommendation method and apparatus, electronic device, and storage medium

Abstract

The application provides a package recommendation method and device, electronic equipment and storage medium, wherein the method comprises: determining the recommendation success rate of each user corresponding to each channel based on the package adaptation degree between the user and the to-be-recommended package, the channel adaptation degree between the user and the channel, the channel package adaptation degree between the channel and the to-be-recommended package, and the influence of the user's social circle in each channel; determining the to-be-marketed user based on the recommendation success rate of each user corresponding to each channel; determining the recommendation channel of each to-be-marketed user based on the channel adaptation degree between each user and each channel; recommending the to-be-recommended package to the corresponding to-be-marketed user based on the recommendation channel of each to-be-marketed user; wherein the package adaptation degree between any user and the to-be-recommended package is determined based on the consumption similarity between the user and the to-be-recommended package and the marginal revenue value of the to-be-recommended package. The application improves the accuracy of package marketing promotion and improves the recommendation effect of the package.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, and in particular to a package recommendation method and device, electronic equipment and storage medium. BACKGROUND

[0002] Telecommunications industry packages have the characteristics of multiple types, rich package content, and rapid iteration. For users, the number of telecommunications packages is large, the bundled content and quota of multiple packages are similar, and the monthly rent is quite different, making it difficult to quickly choose a package that meets user needs from a large number of packages. At the same time, the mismatch between users and packages can cause unnecessary economic losses to users. From the perspective of enterprises, telecommunications companies have difficulty grasping user needs and market trends, and still use extensive advertising methods to promote new packages, causing user aesthetic fatigue and poor actual promotion results. Therefore, how to recommend appropriate packages to appropriate users is increasingly important.

[0003] Current telecommunications package recommendation methods mainly include:

[0004] 1. Current package target user screening method: starting from the package perspective, under the premise of fixed packages, the user group most likely to purchase the package is screened out. Package products have old and new products. For old products, rules / data mining modeling can be used to mine target user groups that may purchase the product. For new products, modeling data is generally started from similar old products to the new product, and the subscription users of similar old products are used as positive samples to screen user groups using rules or data mining methods.

[0005] 2. User package recommendation method: starting from the user perspective, the package most likely to be purchased by the user is screened out. The recommendation of new packages to users is mainly from the perspective of data mining models. Common data mining models include collaborative filtering, K-means clustering, and similarity algorithms.

[0006] However, the above methods have the following defects: traditional marketing activity planning mainly relies on market business experience, which cannot fully consider the rationality of resource allocation, thereby greatly affecting the success rate of marketing activities; target user marketing pre-performance evaluation factors and methods are limited; the user package matching method is relatively limited, resulting in poor package recommendation results. SUMMARY

[0007] The present application provides a package recommendation method, device, electronic equipment and storage medium to solve the defect of poor package recommendation results in the prior art.

[0008] The present application provides a package recommendation method, comprising:

[0009] determine a recommendation success rate of each user corresponding to each channel based on the package adaptation degree between each user and the to-be-recommended package, the channel adaptation degree between each user and each channel, the channel package adaptation degree between each channel and the to-be-recommended package, and the influence of each user in a social circle of each channel;

[0010] determine a to-be-marketed user based on the recommendation success rate of each user corresponding to each channel;

[0011] determine a recommendation channel of each to-be-marketed user based on the channel adaptation degree between each user and each channel;

[0012] recommend the to-be-recommended package to the corresponding to-be-marketed user based on the recommendation channel of each to-be-marketed user;

[0013] The package adaptation degree between any user and a to-be-recommended package is determined based on a consumption similarity between the any user and the to-be-recommended package and a marginal revenue value of the to-be-recommended package.

[0014] According to the package recommendation method provided by the application, the package adaptation degree between any user and a to-be-recommended package is determined based on the following steps:

[0015] determine the marginal revenue value of the to-be-recommended package based on voice, traffic, and monthly fees included in the to-be-recommended package, and T-1-month available voice, available traffic, average revenue per user, voice over package fees, and traffic over package fees; the T is the current month;

[0016] determine the consumption similarity between the any user and the to-be-recommended package based on the average revenue per user, the monthly online traffic per household, and the average call time per household of the any user, the price, the included traffic, and the voice duration of the to-be-recommended package, the number of benefits subscribed by the any user, and the number of benefits included in the to-be-recommended package;

[0017] determine the package adaptation degree between the any user and the to-be-recommended package based on the consumption similarity between the any user and the to-be-recommended package and the marginal revenue value of the to-be-recommended package.

[0018] According to the package recommendation method provided by the application, the channel adaptation degree between each user and each channel is determined based on the following steps:

[0019] determine the weighted channel use times of the any user in the any channel currently; the weighted channel use times of the any user in the any channel currently are determined based on the weighted channel use times of the any user in the any channel last month and the channel use times of the any user in the any channel this month;

[0020] determine a channel adaptation degree between the any user and the any channel based on a weighted channel use number of the any user currently in the any channel.

[0021] According to the package recommendation method provided by the application, the channel package adaptation degree between each channel and the package to be recommended is determined based on the following steps:

[0022] determine a channel user activity value, a user service activity value, a service preference degree and a channel influence degree on service;

[0023] determine a channel package adaptation degree between any channel and the package to be recommended based on the channel user activity value, the user service activity value, the service preference degree and the channel influence degree on service;

[0024] The channel user activity value is a ratio between the number of active users in the any channel who handle the package to be recommended and the number of active users in all channels who handle the package to be recommended.

[0025] The user service activity value is a ratio between the number of active users in the any channel who handle the package to be recommended and the number of active users who handle all services.

[0026] The service preference degree is a ratio between the number of handling of the package to be recommended in the any channel and the overall service handling amount.

[0027] The channel influence degree on service is a ratio between the successful handling amount of the package to be recommended in the any channel and the total successful handling amount of the package to be recommended in all channels.

[0028] According to the package recommendation method provided by the application, the influence of each user in each channel is determined based on the following steps:

[0029] determine a user-level social index of each user; the user-level social index is a sum of social indexes between the corresponding user and each user in the corresponding user's social circle, and the social index between the corresponding user and each user in the corresponding user's social circle is used to represent the social closeness between the corresponding user and each user in the corresponding user's social circle;

[0030] determine the key degree of each user corresponding to each channel based on the user-level social index of each user and the channel adaptation degree between each user and each channel;

[0031] determine the channel key person of the any user corresponding to each channel based on the key degree of each user in the any user's social circle corresponding to each channel;

[0032] Determine the influence circle influence of the any user in each channel based on the influence of the channel key person of the any user in each channel.

[0033] The method for recommending a package provided by the application further comprises the following steps:

[0034] Determine the recommendation success rate of each marketing user in the corresponding recommendation channel;

[0035] Count the number of marketing users whose recommendation success rate in the corresponding recommendation channel is greater than the preset threshold as the number of successful marketing users;

[0036] Determine the success rate of the marketing activity based on the number of successful marketing users and the total number of marketing users.

[0037] The method for recommending a package provided by the application further comprises the following steps:

[0038] Sort each package based on the gap between the average income per user, the monthly online traffic per household and the average call time per household of any user and the price, traffic and voice duration of each package;

[0039] And / or, determine the package adaptation degree between the any user and each package based on the consumption similarity between the any user and each package and the marginal revenue value of each package;

[0040] Determine the pre-recommended package based on the sorting result of each package and / or the package adaptation degree between the any user and each package, and recommend the pre-recommended package to the any user.

[0041] The application further provides a package recommendation device, comprising:

[0042] A user-level success rate calculation unit is configured to determine the recommendation success rate of each user in each channel based on the package adaptation degree between each user and the recommended package, the channel adaptation degree between each user and each channel, the channel package adaptation degree between each channel and the recommended package, and the influence circle influence of each user in each channel.

[0043] A marketing object determination unit is configured to determine the marketing user based on the recommendation success rate of each user in each channel.

[0044] A recommendation channel determination unit is configured to determine the recommendation channel of each marketing user based on the channel adaptation degree between each user and each channel.

[0045] a marketing unit configured to recommend the to-be-recommended package to a corresponding to-be-marketed user based on a recommendation channel of each to-be-marketed user;

[0046] The package adaptation degree between any user and the to-be-recommended package is determined based on a consumption similarity between the any user and the to-be-recommended package and a marginal revenue value of the to-be-recommended package.

[0047] The application further provides an electronic device, including a memory, a processor and a computer program stored in the memory and executable on the processor, and the processor implements the steps of the package recommendation method according to any one of the above.

[0048] The application further provides a non-transitory computer readable storage medium, which stores a computer program, and the computer program is executable on a processor to implement the steps of the package recommendation method according to any one of the above.

[0049] The package recommendation method, device, electronic device and storage medium provided by the application comprehensively consider users, channels and products, respectively quantify each factor that has an influence on users in package marketing, fuse each factor, screen to-be-marketed users with a higher recommendation success rate, screen the most likely recommendation channel for the to-be-marketed users, implement the recommendation of the package, improve the accuracy of package marketing promotion, and improve the recommendation effect of the package. BRIEF DESCRIPTION OF DRAWINGS

[0050] In order to more clearly illustrate the technical solutions of the application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or the prior art description. Obviously, the drawings in the following description are some embodiments of the application, and those skilled in the art can obtain other drawings according to these drawings without creative labor.

[0051] Figure 1 The flowchart of the package recommendation method provided by the application;

[0052] Figure 2 The schematic diagram of the package adaptation degree calculation method provided by the application;

[0053] Figure 3 The modeling schematic diagram of the user channel preference model provided by the application;

[0054] Figure 4 The schematic diagram of the channel package adaptation degree calculation provided by the application;

[0055] Figure 5 A schematic diagram of the circle of influence calculation provided by the present application;

[0056] Figure 6 A structural schematic diagram of the package recommendation device provided by the present application;

[0057] Figure 7 A structural schematic diagram of the electronic device provided by the present application. DETAILED DESCRIPTION

[0058] To make the objectives, technical solutions and advantages of the present application clearer, the technical solutions in the present application will be described below in connection with the drawings in the present application. Obviously, the described embodiments are only some of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without any creative work fall within the protection scope of the present application.

[0059] Figure 1 A flowchart of the package recommendation method provided by the present application, as shown in Figure 1 The method comprises the following steps:

[0060] Step 110, determining the recommendation success rate of each user corresponding to each channel based on the package adaptation degree between each user and the to-be-recommended package, the channel adaptation degree between each user and each channel, the channel package adaptation degree between each channel and the to-be-recommended package, and the circle of influence of each user in each channel;

[0061] Step 120, determining the to-be-marketed user based on the recommendation success rate of each user corresponding to each channel;

[0062] Step 130, determining the recommendation channel of each to-be-marketed user based on the channel adaptation degree between each user and each channel;

[0063] Step 140, recommending the to-be-recommended package to the corresponding to-be-marketed user based on the recommendation channel of each to-be-marketed user;

[0064] The package adaptation degree between any user and the to-be-recommended package is determined based on the consumption similarity between the user and the to-be-recommended package and the marginal revenue value of the to-be-recommended package.

[0065] Specifically, in order to solve the problem of single user package adaptation model factor, not only the matching degree between the recommended package and the user consumption behavior is considered, but also the user's preference for the channel, the promotion adaptation degree of the channel to the new package and the influence degree of the circle of friends to the user are considered, so as to screen out the appropriate marketing user group. Specifically, the package adaptation degree between each user and the recommended package, the channel adaptation degree between each user and each channel, the channel package adaptation degree between each channel and the recommended package, and the influence of the circle of friends of each user in each channel are calculated. Among them, the package adaptation degree can represent the matching degree between the user and the package, the higher the value, the more likely the user is to order the package; the channel adaptation degree can represent the preference degree of the specific user to the specific channel, the higher the value, the more likely the user is to use the channel to handle the business; the channel package adaptation degree can represent the success possibility of promoting the specific package by using the specific channel; the influence of the circle of friends can represent the degree of influence of the specific user in the specific channel by the circle of friends, the higher the value, the more likely the user is to accept the promotion of the channel.

[0066] Here, the package adaptation degree between any user and the recommended package is determined based on the consumption similarity between the user and the recommended package, and the marginal revenue value of the recommended package. Among them, the consumption similarity can represent the similarity between the existing consumption level of the user and the services that the specific package can provide. On the basis of the user's existing voice and traffic content demand, the marginal revenue index is added, which can improve the accuracy of the package adaptation degree.

[0067] In addition, the channel adaptation degree can be predicted by using a machine learning classification model, the channel package adaptation degree can be measured according to the channel user activity index and the business index, and the influence of the circle of friends can be quantified from the influence of the channel key person corresponding to each channel on the user.

[0068] Based on the above package adaptation degree, channel adaptation degree, channel package adaptation degree and influence of the circle of friends, the recommendation success rate of each user corresponding to each channel is determined. Among them, the weighted sum of the above four index values can be obtained to obtain the recommendation success rate of each user corresponding to each channel. For example, the recommendation success rate of any user corresponding to any channel can be calculated by using the following formula:

[0069] 0.7*A1+0.1*B1+0.1*C1+0.1*B2

[0070] Wherein, A1 is the package adaptation degree between the user and the recommended package, B1 is the channel adaptation degree between the user and the channel, C1 is the channel package adaptation degree between the channel and the recommended package, and B2 is the influence of the circle of friends of the user on the channel. The coefficients before each index can be selected according to the actual situation, or can be determined by entropy method, and the embodiments of the present application do not make specific limitation.

[0071] Subsequently, based on the recommendation success rate of each user corresponding to each channel, users with a recommendation success rate higher than a preset threshold can be screened as users to be marketed.

[0072] In addition, according to the channel adaptation degree between each user and each channel, a plurality of channels with a higher channel adaptation degree can be selected as the recommended channels of the entire marketing activity, for example, channels Q1, Q2 and Q3 of Top3. Then, based on the channel adaptation degree between a single user to be marketed and each channel, the channel adaptation degrees of the user to be marketed and each recommended channel are compared, and all users to be marketed are distributed to the above channels to select the optimal recommended channel for each user to be marketed.

[0073] Based on the recommended channel of each user to be marketed, the recommended package can be recommended to the corresponding user to be marketed, so as to realize the promotion and marketing of the recommended package.

[0074] The method provided by the embodiment of the application comprehensively considers users, channels and products, quantifies each factor that has an influence on users in package marketing, fuses each factor, screens users to be marketed with a higher recommendation success rate, and selects the recommended channel most likely to be accepted by the user to be marketed, so as to realize the recommendation of the package, improve the accuracy of package marketing promotion, and improve the recommendation effect of the package.

[0075] Based on the above embodiment, the package adaptation degree between any user and the recommended package is determined based on the following steps:

[0076] Based on the voice, traffic, and monthly fee included in the recommended package, and the available voice, available traffic, average revenue per user, voice over package fee and traffic over package fee of T-1 month, the marginal revenue value of the recommended package is determined; T is the current month;

[0077] Based on the average revenue per user, monthly online traffic per household and average call time per household of the user, the price, included traffic and voice duration of the recommended package, the number of benefits subscribed by the user, and the number of benefits included in the recommended package, the consumption similarity between the user and the recommended package is determined;

[0078] Based on the consumption similarity between the user and the recommended package and the marginal revenue value of the recommended package, the package adaptation degree between the user and the recommended package is determined.

[0079] Specifically, the marginal revenue value of the to-be-recommended package is determined based on voice, flow, and monthly fee included in the to-be-recommended package, and voice available in T-1 month, available flow, average revenue per user in T-1 month, voice overage fee of the to-be-recommended package, and flow overage fee of the to-be-recommended package. For example, the marginal revenue value of the to-be-recommended package can be calculated by using the following formula:

[0080]

[0081] wherein M i is voice included in the to-be-recommended package, m is voice available in T-1 month, D i is flow included in the to-be-recommended package, d is available flow in T-1 month, A i is monthly fee of the to-be-recommended package, A is average revenue per user (ARPU) in T-1 month, mf is voice overage fee in T-1 month, and df is flow overage fee in T-1 month.

[0082] Based on the average revenue per user (ARPU) corresponding to the user, data flow of usage (DOU) and minutes of usage (MOU) per month per household, price, included flow and voice duration of the to-be-recommended package, number of benefits subscribed by the user, and number of benefits included in the to-be-recommended package, the consumption similarity between the user and the to-be-recommended package is determined. The average revenue per user (ARPU) corresponding to the user, data flow of usage (DOU) and minutes of usage (MOU) per month per household can be the average of ARPU, DOU and MOU consumed by the user in the last three months. For example, the consumption similarity between any user and any package can be calculated by using the following formula:

[0083] f = 0.53 * |arpu-A| / A + 0.27 * |dou-D| / D + 0.1 * |mou-M| / M + 0.1 * |n-N| / N

[0084] wherein arpu, dou and mou are the ARPU, DOU and MOU corresponding to the user, A, D and M are the price, included flow and voice duration of the package, n is the number of benefits subscribed by the user, and N is the number of benefits included in the package.

[0085] Based on the consumption similarity between the user and the to-be-recommended package and the marginal revenue value of the to-be-recommended package, the package adaptation degree between the user and the to-be-recommended package can be determined. For example, the package adaptation degree between any user and any package can be calculated by using the following formula:

[0086]

[0087] wherein f if is the consumption similarity between the user and the package min and f max are the minimum and maximum of the consumption similarity between the user and each package, and s is the marginal revenue value of the package.

[0088] Figure 2 The schematic diagram of the package adaptation degree calculation method provided by the embodiment of the application is shown in Figure 2 Before calculating the package adaptation degree between the user and the package, data preprocessing and user screening can be performed. The data preprocessing includes:

[0089] a. Data missing value filling

[0090] Fill 0 for numerical type and "unknown" for character type; fill the available traffic and available voice in the T-1 month package with the DOU in the last 3 months and the MOU in the last 3 months if they are null;

[0091] b. Abnormal value processing

[0092] The data of the user traffic usage and voice usage behavior exceeding the 99.99% quantile point are reduced to the 99.99% quantile point value;

[0093] c. Construction of derived variables

[0094] Calculate the average ARPU in the last 3 months, the average DOU in the last 6 months, the average DOU in the last 3 months, the average MOU in the last 3 months, and the consumption stability.

[0095] d. Calculate the user consumption stability

[0096] The user consumption stability quantification formula is as follows:

[0097]

[0098] AB is the customer consumption stability evaluation result, wherein: a is the T-3 month ARPU; b is the T-2 month ARPU; c is the T-1 month ARPU; and d is the average ARPU in the last 3 months.

[0099] Subsequently, data filtering is performed.

[0100] After filtering the data, for the full amount of user group, first perform reverse elimination:

[0101] a. Eliminate the users who have been 5G package users last month.

[0102] b. Eliminate users who are not allowed to change the main package.

[0103] c. Eliminate users with low monthly traffic (below 5MB) and zero monthly average calling duration in the last 3 months.

[0104] Secondly, the user group is preliminarily positively screened, and users meeting the following conditions are selected:

[0105] (1) Consumption stability <0.15

[0106] (2) The average ARPU in the last three months >=50 yuan

[0107] Among them, the threshold value can be adjusted by oneself.

[0108] Thirdly, ARPU, DOU and MOU are normalized, and the calculation formula is:

[0109]

[0110] Among them, Xmax of ARPU, DOU and MOU is the maximum value of ARPU, DOU and MOU of all users and the larger one of the maximum value of product library price, traffic and voice. Similarly, Xmin of ARPU, DOU and MOU can be obtained. For the user whose DOU / discount ARPU exceeds the average value by 3 times the standard deviation, the following ARPU uses the pre-discount ARPU.

[0111] Based on any of the above embodiments, the channel adaptation degree between each user and each channel is determined based on the following steps:

[0112] Determine the weighted channel use times of any user in any channel at present; the weighted channel use times of the user in any channel at present are determined based on the weighted channel use times of the user in the channel last month and the channel use times of the user in the channel this month;

[0113] Based on the weighted channel use times of the user in the channel at present, determine the channel adaptation degree between the user and the channel.

[0114] Specifically, the channel adaptation degree between each user and each channel can be calculated by constructing a user channel preference model. Figure 3 The modeling schematic diagram of the user channel preference model provided by the embodiment of the present application is shown as Figure 3 The detailed modeling process is as follows:

[0115] First step, determine the input data of the model

[0116] The input data includes data in four dimensions of user basic attributes, customer attributes, channel use and consumption, and the main data is the use data of the user in each channel.

[0117] Second step, establish a channel classification model

[0118] To evaluate the channel preference of the user, a model is built for each channel. The channels include: 10086, short hall, network hall, WeChat hall in the province, APP client in the province, IVR hot spot, business hall, self-service terminal channel, 10085, 1008611, micro store, 10086 WeChat hall, 10086 APP client, etc.

[0119] The sub-model corresponding to each channel includes:

[0120] The weighted channel use times of the user in any channel at present are calculated.

[0121] The weighted channel use times of the user in any channel at present can be obtained by weighted sum of the weighted channel use times of the user in the channel last month and the channel use times of the user in the channel this month. For example, the current weighted channel use times can be calculated by the formula: current weighted channel use times = last month weighted channel use times x 0.4 + this month channel use times x 0.6.

[0122] It should be noted that the weighted channel use times of the starting month are directly taken as the channel use times of the month; the concept of "channel use times" of different channels will be slightly different, which can be counted according to the number of business consultation and handling.

[0123] Then, the channel adaptation degree between the user and the channel can be determined based on the weighted channel use times of the user in the channel at present.

[0124] Specifically, for the user with weighted channel use times less than a certain threshold, the preference is mined by constructing a classification model. The user with obvious preference for the channel (i.e., the weighted channel use times is greater than a certain threshold) is taken as the seed user of the channel, and after removing the channel contact behavior information, the XGB and other binary classification algorithms are used for model training to select the best algorithm, thereby completing the model construction. The constructed classification model is used to predict the user with weighted channel use times less than a specified threshold.

[0125] The evaluation standard of the channel preference degree is constructed, as shown in Table 1, wherein the threshold can be determined according to actual data:

[0126] Table 1

[0127] Channel preference Explanation High Weighted average channel usage >= t1 times Medium t2 <= weighted average channel usage < t1 times Low t3 <= weighted average channel usage < t2 times Potential preference Model prediction

[0128] Subsequently, the channel adaptation degree between the user and the channel is calculated, and the calculation method can be as shown in Table 2:

[0129] Table 2

[0130]

[0131] Based on any of the above embodiments, the channel-package adaptation degree between each channel and the to-be-recommended package is determined based on the following steps:

[0132] Determine the channel user activity value, user service activity value, service preference degree, and channel influence on service degree.

[0133] Based on the channel user activity value, user service activity value, service preference degree, and channel influence on service degree, determine the channel-package adaptation degree between any channel and the to-be-recommended package.

[0134] The channel user activity value is the ratio between the number of active users in the channel that handle the to-be-recommended package and the number of active users in all channels that handle the to-be-recommended package.

[0135] The user service activity value is the ratio between the number of active users in the channel that handle the to-be-recommended package and the number of active users that handle all services.

[0136] The service preference degree is the ratio between the number of to-be-recommended package handling in the channel and the total service handling amount.

[0137] The channel influence on service degree is the ratio between the successful handling amount of the to-be-recommended package in the channel and the total successful handling amount of the to-be-recommended package in all channels.

[0138] Specifically, 6 business categories (package class, traffic package class, voice class, benefit package, family class, and terminal class) and 7 channels (10086, 10085, short hall, micro store, APP client, business hall, and WeChat hall) can be divided. The type of the channel can also be extended to be consistent with the channel type given in the above channel adaptation degree calculation method.

[0139] Figure 4 The schematic diagram of the channel-package adaptation degree calculation provided by the embodiments of the present application is shown in Figure 4 As shown, the business data and channel data are taken, including the number of active users in each channel in the period, the total service handling amount, and the package class service amount, and the following analysis is performed:

[0140] a. Calculate the channel user activity index - channel user activity value b21: the value is the ratio between the number of active users in the channel that handle the to-be-recommended package and the number of active users in all channels that handle the to-be-recommended package.

[0141] b. Calculate the user service activity index - user service activity value b22: the value is the ratio between the number of active users in the channel that handle the to-be-recommended package and the number of active users that handle all services.

[0142] c. Calculate the degree of preference of a certain business - business preference degree b23: the value is the ratio between the handling amount of the recommended package in the channel and the total business handling amount;

[0143] d. Calculate the degree of influence of the channel on the business - channel business influence degree b24: the value is the ratio between the successful handling amount of the recommended package in the channel and the total successful handling amount in all channels.

[0144] Among them, b21 and b22 constitute the active index, and b23 and b24 constitute the business index.

[0145] Based on the channel user activity value, the user business activity value, the business preference degree and the channel business influence degree, the channel package adaptation degree between any channel and the recommended package is determined.

[0146] For example, the channel package adaptation degree can be calculated by the following formula:

[0147] F = 0.42 * (0.55 * b21 + 0.45 * b22) + 0.58 * (0.44 * b23 + 0.56 * b24)

[0148] Among them, the weight of each index can be calculated according to the entropy method.

[0149] Based on any of the above embodiments, the social circle influence of each user in each channel is determined based on the following steps:

[0150] Determine the user-level social index of each user; the user-level social index is the sum of the social indexes between the corresponding user and each user in the social circle of the corresponding user, and the social index between the corresponding user and each user in the social circle of the corresponding user is used to represent the social closeness between the corresponding user and each user in the social circle of the corresponding user;

[0151] Based on the user-level social index of each user and the channel adaptation degree between each user and each channel, determine the key degree of each user corresponding to each channel;

[0152] Based on the key degree of each user in the social circle of any user corresponding to each channel, determine the channel key person of the user corresponding to each channel;

[0153] Based on the influence of the channel key person of the user corresponding to each channel, determine the social circle influence of the user in each channel; the influence of the channel key person is determined based on the relationship between the channel key person and the social opposite user and the key degree of the social opposite user.

[0154] Specifically, Figure 5 The schematic diagram of the social circle influence calculation provided by the embodiment of the present application is as follows: Figure 5As shown, the user-level social index of each user can be determined. The user-level social index is the sum of the social indexes between the corresponding user and each user in the corresponding user's social circle, and the social index between the corresponding user and each user in the corresponding user's social circle is used to represent the social closeness between the corresponding user and each user in the corresponding user's social circle.

[0155] To calculate the social index of a user pair formed by the corresponding user and each user in the corresponding user's social circle, the social frequency of the user pair can be calculated first, and the calculation formula is as shown below:

[0156]

[0157] Where C represents the number of days in the statistical month, D represents the number of communication days, W represents the maximum number of weeks in the statistical month, W represents the number of weeks of communication, P represents the number of statistical months, and P represents the number of communication months. n n n

[0158] Then, the social index of the user pair is calculated, and the formula is as shown below:

[0159]

[0160] The coefficients x, y, and z can be calculated by the analytic hierarchy process.

[0161] The social indexes of all user pairs in the user's social circle are summed to obtain the single user social index, i.e., the user-level social index. The user-level social index is first processed for outliers, and then normalized by the maximum and minimum.

[0162] The user ID is associated with the user channel preference model to determine the single user channel preference, and the channel adaptation degree between each user and each channel is obtained. Since the channel adaptation degree data is divided by channel, the channel key person and its influence are determined by channel in the subsequent process.

[0163] Subsequently, based on the user-level social index of each user and the channel adaptation degree between each user and each channel, the criticality of each user to each channel, i.e., the single user criticality, is determined. For example, the criticality can be calculated by the following formula:

[0164] Criticality = f (social index, channel preference) = x * social index + y * channel preference

[0165] Where the coefficients x and y can be determined by the entropy method.

[0166] ​​​Based on the key degree of each user in the social circle of any user to each channel, if the key degree exceeds a threshold, the user in the social circle is determined as a channel key person of the channel. The influence of the channel key person corresponding to each channel of the user is summed up, and the social circle influence of the user in each channel can be obtained. The influence of the channel key person is determined based on the relationship between the channel key person and the social opposite user and the key degree of the social opposite user.

[0167] Here, the influence of the channel key person on a single user can be calculated, and the influence of each single user is accumulated to obtain the influence of the channel key person. When calculating the influence of a single user, only the opposite user whose key degree is greater than a certain threshold is considered to have an influence on the user, and the close relationship between the user and the opposite user (family members or colleagues) is also considered. Therefore, the influence of the channel key person can be calculated as follows:

[0168] Single user influence: if there is a close relationship, the influence is considered to be the key degree of the opposite user;

[0169] Single user influence: if there is no close relationship, the influence is considered to be the key degree of the opposite user * the social index of the user pair.

[0170] Based on any of the above embodiments, step 140 further comprises:

[0171] Determining the recommendation success rate of each to-be-marketed user corresponding to each recommended channel;

[0172] Counting the number of to-be-marketed users corresponding to each recommended channel whose recommendation success rate is greater than a preset threshold as the number of successful marketing users;

[0173] Based on the number of successful marketing users and the total number of to-be-marketed users, determining the success rate of the marketing activity.

[0174] Specifically, before actually performing marketing promotion, a marketing rehearsal can be performed to estimate the success rate of the entire marketing activity. Specifically, the recommendation success rate of each to-be-marketed user corresponding to each recommended channel can be obtained, and the number of to-be-marketed users corresponding to each recommended channel whose recommendation success rate is greater than a preset threshold (for example, 0.5) is counted as the number of successful marketing users. Subsequently, based on the number of successful marketing users and the total number of to-be-marketed users, the success rate of the marketing activity is determined. For example, the ratio between the number of successful marketing users and the total number of to-be-marketed users can be taken as the success rate of the marketing activity.

[0175] Based on any of the above embodiments, the method further comprises:

[0176] Sort each package based on the gap between the average income per user, the monthly online traffic per household, and the average call time per household of any user and the price, traffic, and voice duration of each package, respectively;

[0177] And / or, determine the package adaptation degree between the user and each package based on the consumption similarity between any user and each package and the marginal revenue value of each package;

[0178] Determine the pre-recommended package based on the sorting result of each package and / or the package adaptation degree between the user and each package, and recommend the pre-recommended package to the user.

[0179] Specifically, the embodiment of the present application also provides a method for identifying a package suitable for a current user. Figure 2 As shown, each package can be sorted based on the gap between the average income per user, the monthly online traffic per household, and the average call time per household of any user and the price, traffic, and voice duration of each package, respectively.

[0180] That is, the distance between the monthly average ARPU (A) of the user and the price of each package (A i ) can be calculated: A i1 -A, A i2 -A,..., A in -A, and sorted in descending order of distance to obtain the serial number rank(A) of each package.

[0181] The distance between the monthly average DOU (D) of the user and the traffic (D i ) contained in each package can be calculated: D i1 -D, D i2 -D,..., D in -D, and sorted in descending order of distance to obtain the serial number rank(D) of each package.

[0182] The distance between the monthly MOU (M) of the user and the voice (M i ) contained in each package can be calculated: M i1 -M, M i2 -M,..., M in -M, and sorted in descending order of distance to obtain the serial number rank(M) of each package.

[0183] Subsequently, the final serial number of each package of the user is calculated to sort each package. The serial number calculation can be shown in the following formula:

[0184] w1*rank(A)+w2*rank(D)+w3*rank(M)

[0185] In addition, the package adaptation degree between the user and each package can be determined based on the consumption similarity between any user and each package and the marginal revenue value of each package. The specific calculation manner of the package adaptation degree is the same as that in the package adaptation degree calculation manner in the above embodiment, and will not be described here.

[0186] The first ranked package can be directly selected as the pre-recommended package based on the ranking result of each package, and recommended to the user. The package with the highest package adaptation degree can be directly recommended as the pre-recommended package to the user. The pre-recommended package can be determined by comprehensively considering the ranking result of each package and the package adaptation degree between the user and each package, and recommended to the user. The present embodiment does not make a specific limitation on this.

[0187] Based on any of the above embodiments, Figure 6 The structure schematic diagram of the package recommendation device provided by the present embodiment is shown in FIG. 6. Figure 6 As shown in the figure, the device comprises a user-level success rate calculation unit 610, a marketing object determination unit 620, a recommendation channel determination unit 630 and a marketing unit 640.

[0188] The user-level success rate calculation unit 610 is configured to determine the recommendation success rate of each user corresponding to each channel based on the package adaptation degree between each user and the to-be-recommended package, the channel adaptation degree between each user and each channel, the channel-package adaptation degree between each channel and the to-be-recommended package, and the influence of the social circle of each user in each channel.

[0189] The marketing object determination unit 620 is configured to determine the to-be-marketed user based on the recommendation success rate of each user corresponding to each channel.

[0190] The recommendation channel determination unit 630 is configured to determine the recommendation channel of each to-be-marketed user based on the channel adaptation degree between each user and each channel.

[0191] The marketing unit 640 is configured to recommend the to-be-recommended package to the corresponding to-be-marketed user based on the recommendation channel of each to-be-marketed user.

[0192] The package adaptation degree between any user and the to-be-recommended package is determined based on the consumption similarity between the user and the to-be-recommended package and the marginal revenue value of the to-be-recommended package.

[0193] The device provided by the embodiment of the application integrates users, channels and products, respectively quantifies each factor that has an influence on users in package marketing, fuses each factor, screens users to be marketed with a higher recommendation success rate, and screens the most likely accepted recommendation channel for the users to be marketed, so as to realize package recommendation, improve the accuracy of package marketing promotion, and improve the recommendation effect of the package.

[0194] Based on any of the above embodiments, the package adaptation degree between any user and the package to be recommended is determined based on the following steps:

[0195] Based on the voice, traffic, and monthly fee included in the package to be recommended, and the available voice, available traffic, average revenue per user, voice over package fee and traffic over package fee in T-1 month, a marginal revenue value of the package to be recommended is determined; T is the current month;

[0196] Based on the average revenue per user, monthly online traffic per household and average call time per household of the user, the price, included traffic and voice duration of the package to be recommended, the number of benefits subscribed by the user, and the number of benefits included in the package to be recommended, a consumption similarity between the user and the package to be recommended is determined;

[0197] Based on the consumption similarity between the user and the package to be recommended and the marginal revenue value of the package to be recommended, a package adaptation degree between the user and the package to be recommended is determined.

[0198] Based on any of the above embodiments, the channel adaptation degree between each user and each channel is determined based on the following steps:

[0199] A weighted channel use frequency of any user in any channel is determined; the weighted channel use frequency of the user in any channel is determined based on the weighted channel use frequency of the user in the channel in the last month and the channel use frequency of the user in the channel in the current month;

[0200] Based on the weighted channel use frequency of the user in the channel, a channel adaptation degree between the user and the channel is determined.

[0201] Based on any of the above embodiments, the channel package adaptation degree between each channel and the package to be recommended is determined based on the following steps:

[0202] A channel user activity value, a user service activity value, a service preference degree and a channel influence degree on service are determined;

[0203] determine a channel package adaptation degree between any channel and the to-be-recommended package based on a channel user activity value, a user business activity value, a business preference degree, and a channel influence degree on the business;

[0204] The channel user activity value is a ratio between the number of active users in the channel that handle the to-be-recommended package business and the number of active users in all channels that handle the to-be-recommended package business.

[0205] The user business activity value is a ratio between the number of active users in the channel that handle the to-be-recommended package business and the number of active users that handle all businesses.

[0206] The business preference degree is a ratio between the number of to-be-recommended package handling in the channel and the overall business handling amount.

[0207] The channel influence degree on the business is a ratio between the successful handling amount of the to-be-recommended package in the channel and the total successful handling amount of the to-be-recommended package in all channels.

[0208] Based on any of the above embodiments, the social circle influence of each user in each channel is determined based on the following steps:

[0209] determine a user-level social index of each user; the user-level social index is a sum of social indexes between the corresponding user and each user in the social circle of the corresponding user, and the social index between the corresponding user and each user in the social circle of the corresponding user is used to represent the social closeness between the corresponding user and each user in the social circle of the corresponding user;

[0210] determine the key degree of each user corresponding to each channel based on the user-level social index of each user and the channel adaptation degree between each user and each channel;

[0211] determine the channel key person of the user corresponding to each channel based on the key degree of each user in the social circle of the user corresponding to each channel;

[0212] determine the social circle influence of the user in each channel based on the influence of the channel key person of the user corresponding to each channel; the influence of the channel key person is determined based on the relationship between the channel key person and the social opposite user and the key degree of the social opposite user.

[0213] Based on any of the above embodiments, the device further comprises a marketing preview unit, configured to:

[0214] determine a recommendation success rate of each to-be-marketed user corresponding to the respective recommended channel;

[0215] count the number of to-be-marketed users corresponding to the respective recommended channel whose recommendation success rate is greater than a preset threshold as the number of successful marketing users;

[0216] The success rate of a marketing campaign is determined based on the number of users successfully marketed and the total number of users to be marketed.

[0217] Based on any of the above embodiments, the device further includes a recommended package determination unit, used for:

[0218] Based on the differences between each user's average income per user, monthly internet data usage per household, and average monthly call time per household, and the prices, data usage, and voice call duration of each package, the packages are ranked.

[0219] And / or, based on the consumption similarity between any user and each package, and the marginal revenue value of each package, determine the package fit between the user and each package;

[0220] Based on the ranking results of each package and / or the compatibility between the user and each package, a pre-recommended package is determined and recommended to the user.

[0221] Figure 7 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 7 As shown, the electronic device may include: a processor 710, a communication interface 720, a memory 730, and a communication bus 740, wherein the processor 710, the communication interface 720, and the memory 730 communicate with each other through the communication bus 740. The processor 710 can call logic instructions in the memory 730 to execute a package recommendation method. This method includes: determining the recommendation success rate for each user across each channel based on the package compatibility between each user and the package to be recommended, the channel compatibility between each user and each channel, the channel-package compatibility between each channel and the package to be recommended, and the influence of each user's social circle on each channel; determining users to be marketed based on the recommendation success rate for each user across each channel; determining the recommendation channel for each user to be marketed based on the channel compatibility between each user and each channel; and recommending the package to be recommended to the corresponding user based on the recommendation channel for each user to be marketed. The package compatibility between any user and the package to be recommended is determined based on the consumption similarity between the user and the package to be recommended, and the marginal revenue value of the package to be recommended.

[0222] Further, the logic instructions in the memory 730 described above can be implemented in the form of software functional units and sold or used as independent products, and can be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the present application essentially or the part that contributes to the prior art or part of the technical solutions can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions to make a computer device (which can be a personal computer, a server, or a network device, etc.) execute all or part of the steps of the methods described in various embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various media that can store program codes.

[0223] In another aspect, the present application also provides a computer program product, which comprises a computer program stored on a non-transitory computer readable storage medium, and the computer program comprises program instructions, when the program instructions are executed by a computer, the computer can execute the package recommendation method provided by the above-mentioned method, and the method comprises: determining a recommendation success rate of each user corresponding to each channel based on a package adaptation degree between each user and a to-be-recommended package, a channel adaptation degree between each user and each channel, a channel package adaptation degree between each channel and the to-be-recommended package, and an influence of a social circle of each user in each channel; determining a to-be-marketed user based on the recommendation success rate of each user corresponding to each channel; determining a recommendation channel of each to-be-marketed user based on the channel adaptation degree between each user and each channel; recommending the to-be-recommended package to the corresponding to-be-marketed user based on the recommendation channel of each to-be-marketed user; wherein the package adaptation degree between any user and the to-be-recommended package is determined based on a consumption similarity between the any user and the to-be-recommended package and a marginal revenue value of the to-be-recommended package.

[0224] In yet another aspect, the present application also provides a non-transitory computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements the above-mentioned package recommendation method, which comprises: determining a recommendation success rate of each user corresponding to each channel based on a package adaptation degree between each user and a to-be-recommended package, a channel adaptation degree between each user and each channel, a channel-package adaptation degree between each channel and the to-be-recommended package, and an influence of a social circle of each user in each channel; determining a to-be-marketed user based on the recommendation success rate of each user corresponding to each channel; determining a recommendation channel of each to-be-marketed user based on the channel adaptation degree between each user and each channel; and recommending the to-be-recommended package to the corresponding to-be-marketed user based on the recommendation channel of each to-be-marketed user; wherein the package adaptation degree between any user and a to-be-recommended package is determined based on a consumption similarity between the any user and the to-be-recommended package and a marginal revenue value of the to-be-recommended package.

[0225] The device embodiments described above are merely illustrative, wherein the units described as separate components can or can not be physically separated, and the components displayed as units can or can not be physical units, i.e., can be located in one place or distributed on multiple network units. Part or all of the modules can be selected to achieve the purpose of the embodiment scheme according to actual needs. Those skilled in the art can understand and implement without creative labor.

[0226] From the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be realized by means of software and the necessary general hardware platform, and of course can also be realized by hardware. Based on such understanding, the above technical solutions can be embodied in the form of a software product, which can be stored in a computer readable storage medium, such as a ROM / RAM, a magnetic disk, an optical disk, etc., and includes a plurality of instructions to make a computer device (which can be a personal computer, a server, or a network device, etc.) execute the methods described in each embodiment or some parts of the embodiments.

[0227] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement to some technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for recommending meal packages, characterized in that, include: Based on the compatibility of each user with the package to be recommended, the compatibility of each user with each channel, the compatibility of each channel with the package to be recommended, and the influence of each user's social circle on each channel, the recommendation success rate of each user for each channel is determined. Based on the recommendation success rate of each user across various channels, identify the users to be marketed to; Based on the channel compatibility between each user and each channel, determine the recommended channels for each user to be marketed. Based on the recommendation channels of each user to be marketed, the recommended packages will be recommended to the corresponding users to be marketed; The suitability between any user and the recommended package is determined based on the consumption similarity between the user and the recommended package, as well as the marginal revenue value of the recommended package. The influence of each user's social circle across various channels is determined based on the following steps: Determine the user-level social index for each user; the user-level social index is the sum of the social indices between the corresponding user and all users in their social circle, and the social index between the corresponding user and all users in their social circle is used to characterize the degree of social closeness between the corresponding user and all users in their social circle; based on the user-level social index of each user and the channel adaptability between each user and each channel, determine the criticality of each user to each channel; based on the criticality of each user's social circle to each channel, determine the channel key person for each channel for each user; Based on the influence of key figures in each channel corresponding to any user, the influence of any user's social circle in each channel is determined; the influence of key figures in each channel is determined based on the relationship between key figures in each channel and users on the other end of the social network, as well as the degree of importance of users on the other end of the social network.

2. The package recommendation method according to claim 1, characterized in that, The compatibility between any user and the recommended plan is determined based on the following steps: Based on the voice, data, and monthly fees included in the package to be recommended, as well as the available voice and data in T-1 month, the average revenue per user, the cost of exceeding the voice and data allowances, the marginal revenue value of the package to be recommended is determined. T refers to the current month; Based on the average income per user, monthly internet traffic per household, and average monthly call time per household for any given user, the price of the package to be recommended, the included data and voice minutes, the number of benefits subscribed by any given user, and the number of benefits included in the package to be recommended, the consumption similarity between any given user and the package to be recommended is determined. Based on the consumption similarity between any user and the recommended package and the marginal revenue value of the recommended package, the package suitability between any user and the recommended package is determined.

3. The package recommendation method according to claim 1, characterized in that, The channel compatibility between each user and each channel is determined based on the following steps: Determine the weighted channel usage count of any user in any channel; the weighted channel usage count of any user in any channel is determined based on the weighted channel usage count of any user in any channel last month and the channel usage count of any user in any channel this month; Based on the weighted number of times any user currently uses any channel on any channel, the channel compatibility between any user and any channel is determined.

4. The package recommendation method according to claim 1, characterized in that, The compatibility between each channel and the recommended package is determined based on the following steps: Determine the channel's user activity level, user business activity level, business preference level, and the channel's impact on the business; Based on the channel user activity value, the user business activity value, the business preference value, and the influence of the channel on the business, determine the channel package compatibility between any channel and the package to be recommended; Wherein, the channel user activity value is the ratio between the number of active users who have subscribed to the recommended package service in any channel and the number of active users who have subscribed to the recommended package service in all channels; The user service activity value is the ratio between the number of active users who have subscribed to the recommended package service in any channel and the number of active users who have subscribed to all services. The degree of business preference is the ratio between the number of transactions for the recommended package within any channel and the total number of transactions. The degree of influence of a channel on business is the ratio of the number of successful transactions of the recommended package on any one channel to the total number of successful transactions across all channels.

5. The package recommendation method according to any one of claims 1 to 4, characterized in that, The step of recommending the package to the corresponding user based on the recommendation channel of each user to be marketed also includes: Determine the success rate of each user to be marketed to through their respective recommendation channels; The number of users to be marketed who have a recommendation success rate greater than a preset threshold for their respective recommendation channels is counted as the number of successfully marketed users. The success rate of the marketing campaign is determined based on the number of successfully marketed users and the total number of users to be marketed.

6. The package recommendation method according to any one of claims 1 to 4, characterized in that, Also includes: Based on the differences between each user's average income per user, monthly internet data usage per household, and average monthly call time per household, and the prices, data usage, and voice call duration of each package, the packages are ranked. And / or, based on the consumption similarity between any user and each package, and the marginal revenue value of each package, determine the package fit between any user and each package; Based on the ranking results of each package and / or the package compatibility between any user and each package, a pre-recommended package is determined and recommended to any user.

7. A package recommendation device, characterized in that, include: The user-level success rate calculation unit is used to determine the recommendation success rate of each user for each channel based on the package compatibility between each user and the package to be recommended, the channel compatibility between each user and each channel, the channel package compatibility between each channel and the package to be recommended, and the influence of each user's social circle on each channel. The target audience identification unit is used to identify users to be marketed based on the recommendation success rate of each user across various channels. The recommendation channel determination unit is used to determine the recommended channels for each user to be marketed based on the channel compatibility between each user and each channel. The marketing unit is used to recommend the package to the corresponding user based on the recommendation channel of each user to be marketed; The suitability between any user and the recommended package is determined based on the consumption similarity between the user and the recommended package, as well as the marginal revenue value of the recommended package. The influence of each user's social circle across various channels is determined based on the following methods: Determine the user-level social index for each user; the user-level social index is the sum of the social indices between the corresponding user and all users in their social circle, and the social index between the corresponding user and all users in their social circle is used to characterize the degree of social closeness between the corresponding user and all users in their social circle; based on the user-level social index of each user and the channel adaptability between each user and each channel, determine the criticality of each user to each channel; based on the criticality of each user's social circle to each channel, determine the channel key person for each channel for each user; Based on the influence of key figures in each channel corresponding to any user, the influence of any user's social circle in each channel is determined; the influence of key figures in each channel is determined based on the relationship between key figures in each channel and users on the other end of the social network, as well as the degree of importance of users on the other end of the social network.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps of the package recommendation method as described in any one of claims 1 to 6.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the package recommendation method as described in any one of claims 1 to 6.

Images