WO2020258773A1 - Procédé, appareil et dispositif de détermination d'un groupe d'utilisateurs d'envoi et support de stockage - Google Patents

Procédé, appareil et dispositif de détermination d'un groupe d'utilisateurs d'envoi et support de stockage Download PDF

Info

Publication number
WO2020258773A1
WO2020258773A1 PCT/CN2019/126717 CN2019126717W WO2020258773A1 WO 2020258773 A1 WO2020258773 A1 WO 2020258773A1 CN 2019126717 W CN2019126717 W CN 2019126717W WO 2020258773 A1 WO2020258773 A1 WO 2020258773A1
Authority
WO
WIPO (PCT)
Prior art keywords
push
user
user group
message
arrival rate
Prior art date
Application number
PCT/CN2019/126717
Other languages
English (en)
Chinese (zh)
Inventor
徐骄
Original Assignee
广州视源电子科技股份有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 广州视源电子科技股份有限公司 filed Critical 广州视源电子科技股份有限公司
Publication of WO2020258773A1 publication Critical patent/WO2020258773A1/fr

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q30/00—Commerce
    • G06Q30/02—Marketing; Price estimation or determination; Fundraising
    • G06Q30/0201—Market modelling; Market analysis; Collecting market data
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q30/00—Commerce
    • G06Q30/02—Marketing; Price estimation or determination; Fundraising
    • G06Q30/0241—Advertisements
    • G06Q30/0277—Online advertisement

Definitions

  • the present disclosure relates to the field of information push technology, for example, to a method, device, equipment, and storage medium for determining a push user group.
  • the push target is usually determined based on the historical push messages reaching the application and the user's click situation. If the application does not have historical push messages, all users will be the push targets.
  • the above method for determining the push object lacks rationality. For example, to determine the push target only based on historical push messages, new users using the application will be ignored. For another example, using all users as the push target will increase the burden of the push platform. At the same time, users who are disgusted with pushing messages will reduce the experience, which may cause serious consequences of losing users of the application.
  • the present disclosure provides a method, device, equipment, and storage medium for determining a push user group, so as to solve the technical problem of lack of rationality in the determination of push objects in related technologies.
  • embodiments of the present disclosure provide a method for determining a push user group, including:
  • the first target user group is a user group that has not accepted the push behavior.
  • the obtaining the user level of each user in the first target user group includes:
  • the identifying the attribute tag to determine the user level of the corresponding user includes:
  • the attribute label is identified by a user classification model to determine the user level of the corresponding user, and the user classification model is obtained by training the attribute label and user level of each user in the first reference user group.
  • the method further includes:
  • the message push data of each user in the second target user group is recorded, and the message push data includes: the number of first message pushes and/or the number of message clicks.
  • it further includes:
  • the user tags include at least one of attribute tags, operation data, operation derived data, and user levels.
  • the third target user group is User groups who have accepted the push behavior;
  • the obtaining the second arrival rate estimation value of each user according to the user tag and corresponding message push data includes:
  • the arrival rate estimation model is used to identify the user tags and the corresponding message push data to obtain the second arrival rate estimation value of each user.
  • the arrival rate estimation model trains the second reference user group The user tag, message push data, and the estimated value of the arrival rate of each user are obtained.
  • the selecting a first set number of users from the first target user group according to the first arrival rate estimation value to form a second target user group includes:
  • a first set number of users are selected from the first target user group to form a second target user group.
  • the first set number of users are selected from the first target user group to form a second target user group according to the number of pushes of the second message and the first estimated arrival rate include:
  • the pushing a message to the second target user group includes:
  • the current push method includes pop-up push
  • the arrival rate prediction model includes a pop-up arrival rate prediction model
  • the current push method includes information push
  • the arrival rate estimation model includes an information arrival rate estimation model
  • embodiments of the present disclosure also provide a device for determining a push user group, including:
  • the obtaining module is used to obtain the user level and estimated coefficient of each user in the first target user group, the estimated coefficient is determined by the operation data of the corresponding user, and the operation data is generated when the user operates the application program The data;
  • a calculation module configured to calculate an estimated value of the first arrival rate of each user according to the user level and the estimated coefficient
  • a determining module configured to select a first set number of users in the first target user group according to the first estimated arrival rate to form a second target user group;
  • the push module is used to push messages to the second target user group.
  • embodiments of the present disclosure also provide a device for determining a push user group, including:
  • One or more processors are One or more processors;
  • Memory used to store one or more programs
  • the one or more processors When the one or more programs are executed by the one or more processors, the one or more processors implement the method for determining the push user group as described in the first aspect.
  • the embodiments of the present disclosure also provide a storage medium containing computer-executable instructions, when the computer-executable instructions are executed by a computer processor, the computer-executable instructions are used to perform the push user group determination as described in the first aspect. method.
  • the above method, device, equipment and storage medium for determining the push user group calculate the estimated value of the first arrival rate of each user by obtaining the user level and estimation coefficient of each user in the first target user group, and calculate the estimated value of each user's first arrival rate according to the first arrival rate
  • the estimated value determines the second target user group that can push the message, and then pushes the message to the second target user group.
  • the technical means can be used to select a reasonable user group according to the user’s level and operation data when pushing the message. For user groups who have not pushed messages before, there is no need to push all users, which prevents waste of resources and at the same time improves the input-output ratio of push messages.
  • FIG. 1 is a flowchart of a method for determining a push user group according to Embodiment 1 of the present disclosure
  • FIG. 2 is a flowchart of a method for determining a push user group according to Embodiment 2 of the disclosure
  • FIG. 3 is a schematic diagram of user classification model training provided in Embodiment 2 of the disclosure.
  • FIG. 5 is a schematic structural diagram of an apparatus for determining a push user group provided by Embodiment 4 of the present disclosure
  • FIG. 6 is a schematic structural diagram of a device for determining a push user group provided by Embodiment 5 of the disclosure.
  • FIG. 1 is a flowchart of a method for determining a push user group provided in Embodiment 1 of the present disclosure.
  • the method for determining the push user group provided in the embodiment may be executed by the device that determines the push user group.
  • the device for determining the push user group may be implemented by software and/or hardware.
  • the device for determining the push user group may be two Or multiple physical entities, or one physical entity.
  • the device for determining the push user group may be a computer, a mobile phone, a tablet, or a smart interactive tablet.
  • the device that determines the user group to push can be understood as the business background of the set application, or a message push platform that can push messages to the set application, or a device that integrates the business background and the message push platform , That is, it is determined that the device pushing the user group has the server function.
  • the specific type of the setting application is not limited in the embodiment, which can be any one or several associated applications.
  • the device with the setting application and the device that determines the user group are different physical The physical device, usually, the device installed with the setting application is an electronic device used by the user.
  • the device integration business backend and the message push platform of the push user group are determined as an example.
  • the device that is determined to push the user group can generate a push message according to actual needs, and push the message to the set user or user group. It is understandable that pushing a message to a user refers to pushing a message to a set application used by the user. In the embodiment, the behavior of pushing a message is recorded as a pushing behavior.
  • the method for determining a push user group includes:
  • Step 110 Obtain the user level and estimated coefficient of each user in the first target user group.
  • the estimated coefficient is determined by the operation data of the corresponding user, and the operation data is the data generated when the user operates and sets the application.
  • the first target user group is a user group using the setting application, which may include all users using the setting application, or a user group using the setting application within a set time period, or It includes users who meet the set conditions.
  • the first target user group is a user group that has not accepted the push behavior. For example, among all users who set the application, select all users who have not accepted the push behavior or a set number of users to form the first target user group. For another example, when a message is pushed for the first time within a set time after the set application goes online, all users currently using the set application or a set number of users are taken as the first target user group.
  • the users in the first target user group may be registered users or unregistered users who use the setting application.
  • the device that determines the user group can be determined by setting the application to determine the user's device model, the user's device price, the user's device version, the setting of the application download channel, the setting of the application version number, and the settings. Specify at least one item of attribute data such as application activation channels. You can also set the application to determine the user's operating behavior.
  • the operation behavior includes at least one of user login time, user usage time, user payment amount and payment time, user online time, and user browsing time.
  • the device that determines the push user group can also determine at least one attribute data such as the user's login name or nickname, gender, age, and application login channel through the setting application.
  • the device that determines the push user group can also determine the attribute data such as the login address by setting the application.
  • the device that determines the push user group can obtain attribute data when the user downloads and uses the setting application and/or registers.
  • each user in the first target user group has a corresponding user level.
  • the user level can indicate the user's popularity and proficiency in setting the application.
  • the level type included in the user level can be set in accordance with the actual situation.
  • the set user levels include at least: advanced users, intermediate users, primary users, and lost users.
  • the user level determination method can be set according to actual conditions. For example, the user level is determined through the user's operation behavior, and another example is the user level determined through the user's attribute data.
  • the restriction conditions corresponding to different user levels it is possible to pre-set the restriction conditions corresponding to different user levels and confirm the restriction conditions satisfied by the user's operation behavior and/or attribute data, or to establish a set in advance, which includes certain A number of operation behaviors and/or attribute data and corresponding user levels, after which a model is constructed and the above data is trained, in order to identify the corresponding behavior data and/or attribute data through the model in the subsequent process, and then obtain the user level.
  • the user level corresponding to each user is not fixed data, which can be changed according to the actual usage of the user.
  • each user has a corresponding estimation coefficient.
  • the estimated coefficient can be understood as a weight used to describe the degree of user activity.
  • the estimation system is determined by operating data of the corresponding user.
  • the operation data is the data generated when the user operates the application program, that is, the operation data is obtained through the operation behavior. For example, if the operation behavior is that the user logs in to the setting application, the corresponding operation data includes: login time and the number of logins within the set time period.
  • the weights corresponding to different operation data are preset, and then the weights corresponding to the user are determined in combination with the user's operation behavior, and the weights are used as the estimated coefficient of the user.
  • the user operation data is that there are payment records in a set time period (such as the past three days), and the corresponding estimation coefficient is 0.8.
  • the user operation data is that there are login records every day within a set number of days (such as the past three days), and the corresponding estimated coefficient is 0.5.
  • the user operation data is a set number of days (such as nearly three days)
  • the estimated coefficient that is, the estimated coefficient of the payment record in the set time period (such as the past three days) is used as the user's estimated coefficient. It is understandable that the estimated coefficient corresponding to each user is not fixed data, and it can be changed according to the actual usage of the user.
  • Step 120 Calculate the estimated value of the first arrival rate of each user according to the user level and the estimated coefficient.
  • the first estimated arrival rate indicates the message arrival probability when the message is pushed to the user.
  • the higher the estimated first arrival rate the higher the success rate of pushing messages to users.
  • the estimated value of the first arrival rate is determined by the user level and the estimated coefficient. In one embodiment, different values are set for different user levels in advance. Generally speaking, the higher the user level, the lower the corresponding value can be. The advantage of this is that when you choose to push the user group later, you can select some users with general or low activity to increase the user's activity by pushing messages.
  • the calculation method of the user level and the estimation coefficient is not limited. For example, the product of the value corresponding to the user level and the prediction coefficient is used as the first arrival rate prediction value. For another example, the user level and the weight value of the estimated coefficient are respectively set, and then the first arrival rate estimated value is determined by setting a binary linear equation.
  • user levels include advanced users, intermediate users, primary users, and process users.
  • Set a user as an intermediate user, and its corresponding estimated coefficient is 0.5, then the estimated value of its first arrival rate is: 2 ⁇ 0.5 1.
  • the probability of middle-level users being pushed is higher than the probability of being pushed by advanced users. In this way, the interest of intermediate users can be increased through message push, and the probability of intermediate users becoming advanced users can be increased.
  • Step 130 Select a first set number of users from the first target user group according to the first estimated value of arrival rate to form a second target user group.
  • the second target user group is a user group for message push, that is, the push object.
  • the second target user group is a subset of the first target user group.
  • the first arrival rate estimate of each user in the first target user group may be sorted in descending order, and then according to the sorting result and the first A set number to confirm the second target user group.
  • the number of users in the second target user group ie, the first set number
  • the specific value size may also be flexibly set according to actual conditions.
  • the 300 users with the highest estimated first arrival rate are selected to form the second target user group.
  • the 30% users with the highest estimated first arrival rate are selected from the ranking results to form the second target user group.
  • the first set number can be a larger value, and if the number of users in the first target user group is small, the first set number can be set to a smaller value.
  • the number of users When the number of users is large, perform multiple comprehensive analysis of the message push results and optimize the push content or push quantity. You can also decentralize access when necessary to avoid centralized server access and excessive load.
  • a certain number of users can also be randomly selected from the remaining users in the first target user group , Together to form the second target user group.
  • the number of users selected in the two modes may be the same or different. For example, select the 10% users with the highest estimated first arrival rate, and randomly select 5% of the remaining users to form the second target user group together.
  • Step 140 Push messages to the second target user group.
  • the message to be pushed is edited, the communication address of the second target user group is obtained, and the message is pushed to the second target user group.
  • the push method is predetermined, and the message push is performed according to the push method.
  • the push mode refers to the display mode of the message in the setting application.
  • the push mode includes but is not limited to: pop-up push and/or information push.
  • Pop-up push refers to displaying messages in the setting application by displaying pop-up windows.
  • Information push refers to the display of push messages in the form of notifications. When the user clicks on the notification, the message is displayed on the message display interface.
  • the setting application can record whether the user clicks on the message, and report to the device that determines the push user group, and the device that determines the push user group can regard whether the user clicks on the message as the follow-up push message. Reference data.
  • the user group that pushes the message again can be determined by combining parameters such as the operation behavior of each user, the user level, the number of message pushes, and the number of user clicks.
  • Example 1 Set the application to be an application with a set duration (for example, three months after being launched), and the set application has not yet been pushed to the user.
  • the device that determines the push user group can regard all current users as the first target user group, and obtain the user level and corresponding estimated coefficient of each user in the first target user group, and then obtain the first arrival rate of each user
  • the estimated value, the second target user group is selected by the first arrival rate estimated value, and the message is pushed to the second target user group.
  • Example 2 When determining the device that pushes the user group for message push, first, obtain all users who have not pushed the message to form the first target user group, and obtain the user level of each user in the first target user group and the corresponding estimated coefficient , And then obtain the first estimated value of arrival rate of each user, select the second target user group through the first estimated value of arrival rate, and push the message to the second target user group.
  • the first estimated value of arrival rate of each user is calculated by obtaining the user level and estimated coefficient of each user in the first target user group, and the second target user who can push the message is determined according to the first estimated value of arrival rate Groups, and then the technical means of pushing messages to the second target user group.
  • pushing messages a reasonable user group can be selected according to the user’s level and operating data. For user groups that have not pushed messages, it is not necessary to perform information on all users. Pushing prevents waste of resources and at the same time improves the input-output ratio of push messages.
  • Fig. 2 is a flowchart of a method for determining a push user group provided in the second embodiment of the disclosure.
  • the method for determining the push user group is to embody the above method for determining the push user group.
  • the first target user group is set as a user group that has not accepted the push behavior. That is, it is determined that the device pushing the user group has not pushed messages to users in the first target user group.
  • the method for determining a push user group includes:
  • Step 201 Obtain attribute tags of users in the first target user group.
  • the attribute tag it is set to determine the user level of the user in the first target user group through the attribute tag, and at the same time, the user targeted by the set application is restricted to be a registered user.
  • the attribute label is determined according to the attribute data.
  • the set attribute tags include: gender, user device model, user device price, user device version, set application download channel, set application login channel, set application activation channel, and Set at least one of the version numbers of the application.
  • the attribute tag includes all the above contents.
  • the setting application can obtain the user's device model, the user's device version, set the application download channel, gender, set the application login channel, and set the application activation Channels and set the version number of the application, and report the above data to the device that determines the push user group.
  • the specific method of setting the application to obtain data and determining the price of the user's use of the device is not limited.
  • Step 202 Identify the attribute tag to determine the user level of the corresponding user.
  • attribute tags corresponding to different user levels are preset, and after the attribute tags are obtained, the corresponding user level can be determined according to the corresponding relationship. Or, the attribute tags are identified through the user classification model to determine the corresponding user level.
  • the user level is determined through the user classification model as an example for description.
  • the step is: identifying the attribute tag through the user classification model to determine the user level of the corresponding user.
  • the user classification model is obtained by training the attribute tags and user levels of each user in the first reference user group.
  • the first reference user group refers to a set of users with specific attribute tags and user levels.
  • the content of the first reference user group may be manually determined by the background personnel who set the application, or the known attribute tags and corresponding user levels in other applications may be obtained, and the first reference user group may be constructed.
  • the attribute tag of the first reference user group is used as input, and the user level is used as output, and machine learning is used for training to obtain a user classification model.
  • the specific machine learning method can be set according to the actual situation.
  • LightGBM is taken as an example for description.
  • LightGBM is a machine learning model that uses the histogram algorithm. LightGBM uses unsupervised attribute tags and adds functions to the tags to generate a user classification model to predict user levels.
  • FIG. 3 is a schematic diagram of user classification model training provided in Embodiment 2 of the disclosure. That is, LightGBM is used to train attribute tags of known user levels, and then a user classification model is obtained.
  • the advantage of using LightGBM is that the training speed is fast, and discrete features can be directly trained, such as directly training users to use the device model.
  • the attribute tags of the users in the first target user group can be identified, and the corresponding user level can be output. It is understandable that the user classification model can be regularly trained to ensure classification accuracy.
  • Step 203 Obtain the estimated coefficient of each user in the first target user group.
  • the estimated coefficient is determined by the operating data of the corresponding user, and the operating data is the data generated when the user operates the application program.
  • step 203 can also be executed before step 201, or simultaneously with step 201 and step 202.
  • Step 204 Calculate the estimated value of the first arrival rate of each user according to the user level and the estimated coefficient.
  • the set user level is represented by Modelbase
  • the estimated coefficient is represented by ⁇
  • the estimated value of the first arrival rate is represented by y
  • y Modelbase* ⁇ .
  • the user classification model can directly output the value of the user level.
  • Step 205 Determine the second message push times of the set application.
  • the second number of message pushes refers to the number of times the set application program performs message pushes. If the number of message pushes is zero, it means that the set application has not pushed messages to the user. Generally speaking, when the application is set to push a message once, the second message push times increase by 1.
  • Step 206 According to the second message push times and the first estimated arrival rate, select a first set number of users from the first target user group to form a second target user group.
  • this step is set in the embodiment to include steps 2061 to 2063:
  • Step 2061 confirm whether the second message push times are within the set times.
  • step 2062 is executed.
  • step 2063 is executed.
  • message push can be divided into: unpush stage, push initial stage and push mature stage.
  • the non-push stage means that no message push has been carried out, and the initial stage of push means that there are fewer message pushes.
  • the push maturity stage refers to the push of more messages.
  • the user push surface is fixed.
  • the user group selection methods corresponding to different stages are different. Therefore, each stage can be distinguished by setting the range of times to help determine the screening method of the user group.
  • the minimum value of the set frequency range is generally 1, and the maximum value can be set in conjunction with the daily active users and total users of the set application. For example, when setting the application to target a larger number of user groups, the maximum value can be larger.
  • the second message is pushed When the number of times the second message is pushed is within the set number of times, it indicates that it is in the initial stage of pushing. When the number of times the message is pushed is less than the set number of times, it means it is in the non-push stage. When the number of times the message is pushed is greater than the set number of times, it indicates that the push is in the mature stage.
  • Step 2062 according to the order of the first estimated arrival rate from high to low, select the first number of users in the first target user group to form the first subgroup, and randomly select the second number of users from the remaining users to form the first subgroup
  • the second subgroup, the first subgroup and the second subgroup form a second target user group.
  • the specific values of the first quantity and the second quantity can be set according to actual conditions. Generally speaking, the value corresponding to the first number is greater than the value corresponding to the second number.
  • the click-through rate of the push messages can be guaranteed, that is, the input-output ratio can be guaranteed.
  • randomly selecting the second number of users to form the second subgroup can ensure the expansion of user push coverage.
  • a part of the user groups that have accepted the push behavior may also be selected as the user group for message push to ensure the rationality of the push users.
  • Step 2063 According to the order of the first estimated arrival rate from high to low, select a third number of users from the first target user group to form a second target user group.
  • the embodiment sets that whether it is in the non-push stage or in the push mature stage, for the first target user group, the set number of users with the highest first arrival rate estimate is selected to form the second target user group.
  • the third quantity corresponding to the non-push stage and the third quantity corresponding to the push mature stage may be different.
  • the users who set the application for are users who have not accepted the push behavior. At this time, you can only choose to push users in the first target user group. Therefore, the third number can be set to one A larger value to ensure the number of pushes.
  • the application is set to target two types of user groups: one is the user group that has accepted the push behavior, and the other is the user group that is new and has not yet accepted the push behavior. That is, in addition to selecting push users in the first target user group, you can also choose to push users in the user group that has accepted the push behavior.
  • the third number can be set to a small value to ensure that even in the mature stage of the push, the new user can still be used as the push user, thereby increasing the activity of the new user.
  • the second target user group includes the first subgroup and the second subgroup.
  • 20% of the users who have accepted the push behavior are selected as push targets in accordance with the set rules. If the number of times of pushing the second message is higher than the set number of times, in the first target user group, in the order of the first arrival rate estimation value from high to low, 10% of users are selected to form the second target user group. At the same time, 20% of the users who have accepted the push behavior are selected as push targets in accordance with the set rules.
  • Step 207 Push messages to the second target user group.
  • this step includes step 2071 to step 2072:
  • Step 2071 determine the current push method, the current push method includes: pop-up push and/or information push.
  • the current push method includes pop-up push and information push as an example for description.
  • the specific means for determining the current push mode is not limited in the embodiment.
  • the push mode flag is preset, and different values in the flag represent different push methods. When pushing messages, read the flag to determine the current push mode.
  • Step 2072 push the message according to the current push mode.
  • the message is edited according to different push methods and sent to the device where the setting application is located, so that when the user uses the setting application, the message is pushed according to the corresponding push method.
  • Step 208 Record the message push data of each user in the second target user group.
  • the message push data includes: the number of first message pushes and/or the number of message clicks.
  • the first message push times refers to the number of times the user has been pushed.
  • the number of push clicks refers to the number of times a user clicks on a push message. Generally, the number of first message pushes and the number of message clicks of each user in the first target user group are both zero.
  • the number of times of pushing the first message corresponding to the user is increased by one.
  • the application is set to display the push message to the user, it can be confirmed whether the user clicks on the message, and feedback is given to the device that determines the push user group.
  • the number of clicks on the message corresponding to the user increases by 1. It is understandable that after each message is pushed, the message push data of the user who is pushed will be updated. The advantage of recording the message push data is to facilitate the understanding of the arrival of the message and the situation of the message being watched, and then determine the input-output ratio of the push message. At the same time, the message push data can also be used as reference data for selecting push user groups.
  • each user includes two types of message push data. One type corresponds to pop-up push, and the other type is used for message push.
  • step 209-step 212 in addition to pushing messages to user groups that have not accepted the push behavior, if the set application has already generated push behaviors, you also need to select push among the user groups that have accepted the push behavior Objects to ensure the rationality of the selection of push objects.
  • step 209-step 212 taking the third target user group as the user group in the setting application that has accepted the push behavior as an example, the process of selecting the push object in the third target user group is described.
  • Step 209 Obtain user tags and message push data of users in the third target user group.
  • the user tag includes at least one of attribute tags, operation data, operation derived data, and user levels.
  • the setting operation data at least includes: daily open period and daily online duration.
  • the daily open period can record the user's habitual use time.
  • the daily online time can determine the user's active program and the degree of dependence on the set application.
  • the setting application can generate a user operation log and report it to the device that determines the push user group, and the device that determines the push user group obtains the operation data through the operation log.
  • operation derived data refers to regular data obtained after summarizing and summarizing user operation data.
  • the setting operation derivative data includes at least: a time period when the user frequently pays and a time period when the user frequently browses.
  • the time period when users often pay is determined according to the user's payment behavior and payment time.
  • the user browsing time period is determined by the user's daily open period and daily online time.
  • the user's operation log also includes the user's payment behavior and the user's payment time.
  • the user's operating habits can be inferred by operating derived data.
  • the user level can be obtained through a user classification model, and can also be determined according to the user's operation behavior. For example, if a user's daily online duration exceeds the duration threshold and the payment frequency is high, then the user can be determined as an advanced user. For another example, if a user has not used the setting application in the near future, the user can be determined as a lost user.
  • the method of obtaining the attribute tag is the same as the method of obtaining the attribute tag in step 201, and will not be repeated here.
  • the message push data can be updated after each message push.
  • Step 210 Obtain the second arrival rate estimation value of each user according to the user tag and corresponding message push data.
  • the second estimated value of arrival rate has the same meaning as the first estimated value of arrival rate.
  • the arrival rate estimation model when determining the second arrival rate estimation value according to the user tag and corresponding message push data, can be used to determine it, or alternatively, different second arrival rate estimation values and users can be preset
  • the correspondence between the label and the corresponding message push data is determined according to the correspondence.
  • the use of the arrival rate prediction model is taken as an example for description.
  • this step includes: using the arrival rate estimation model to identify the user tag and the corresponding message push data to obtain the second arrival rate estimation value of each user.
  • the arrival rate estimation model is obtained by training the user tags, message push data, and arrival rate estimation values of users in the second reference user group, and is used to predict the message arrival probability.
  • the second reference user group refers to users whose second arrival rate estimation value is clear.
  • the second reference user group can be manually set by the background personnel who set the application, or obtain the user tags, message push data, and second arrival rate estimation value of each user in other applications, and form the second reference user group. 2.
  • Reference user group In one embodiment, the user tags and message push data of the second reference user group are used as input, and the second arrival rate estimation value is used as output, and machine learning is used for training to obtain the arrival rate estimation model.
  • the specific machine learning method can be based on actual conditions.
  • LightGBM is still used as an example for description.
  • the arrival rate estimation model After the training of the arrival rate estimation model is completed, the user tags and message push data of users in the third target user group can be identified, and the corresponding second arrival rate estimation value can be output. It is understandable that the arrival rate estimation model can be regularly trained to ensure accuracy.
  • the current push method since the current push method includes pop-up push and information push, and the message push data is divided into data corresponding to the pop-up push and data corresponding to the information push, then when constructing the arrival rate estimation model , You can build a pop-up window arrival rate prediction model for pop-up window push, and build an information arrival rate prediction model for information push. That is, the current push method includes pop-up push, and the arrival rate prediction model includes the pop-up arrival rate prediction model. The current push method includes information push, and the arrival rate estimation model includes the information arrival rate estimation model.
  • the input when constructing the pop-up window arrival rate prediction model, includes the user tag and message push data of each user corresponding to the pop-up window push in the second reference user group.
  • the input when constructing the information arrival rate estimation model, includes the user tag and message push data of each user corresponding to the information push in the second reference user group.
  • after constructing the pop-up window arrival rate prediction model and the information arrival rate prediction model first determine the current push method and the third target user group corresponding to the current push method, and then select the corresponding arrival rate prediction model Get the estimated value of the second arrival rate.
  • Step 211 Select a second set number of users in the third target user group according to the second estimated value of arrival rate to form a fourth target user group.
  • the specific value of the second set number can be set according to actual conditions, and the fourth target user group is the determined push target.
  • the second set number of users when determining the fourth target user group, may be selected in descending order of the second estimated arrival rate, or the second estimated arrival rate may be selected Users who are higher than the reference arrival value, or, the second set number includes the first sub-number and the second sub-number, and the first sub-number with the highest estimated second arrival rate is selected in the third target user group Users, and randomly select a second sub-number of users from the remaining users in the third target user group.
  • Step 212 Push messages to the fourth target user group.
  • the method of pushing messages to the fourth target user group is the same as that of pushing messages to the second target user group, and will not be repeated here.
  • the corresponding message push data is updated synchronously to ensure the real-time and accuracy of the message push data.
  • the first message push is performed.
  • all users targeted by the setting application are regarded as the first target user group, and the attribute tags of each user in the first target user group are identified through the user classification model to determine the user level of each user.
  • the estimated value of the first arrival rate is determined based on the estimated coefficient and user level determined by each user during the operation.
  • the 30% users with the highest estimated first arrival rate are selected to form the second target user group, and messages are pushed to the second target user group. After the push is completed, record the message push data of the second target user.
  • the user classification model is used to identify the attribute tags of each user to determine the corresponding user level, and the first arrival of each user is determined based on the user level and the corresponding estimated coefficient According to the estimated value of the first arrival rate and the current number of pushes, the second target user group is selected as the push target from the first target user group.
  • the user classification model is used to identify the attribute tags of each user to determine the corresponding user level, and the first arrival of each user is determined based on the user level and the corresponding estimated coefficient
  • the second target user group is selected as the push target from the first target user group.
  • Identify the user tag and message push data of each user through the arrival rate estimation model to determine the second arrival rate estimated value of each user.
  • the third target user who has accepted the promotion behavior Select the fourth target user group in the group as the push target, and push the message for the push target.
  • the technical means of recording the message push data of each user can select a reasonable user group when pushing the message.
  • the push behavior user group and the user group that have accepted the push behavior adopt different selection methods, which can maintain existing push users and find potential push users, prevent waste of resources, and increase the input-output ratio of push messages.
  • FIG. 4 is a flowchart of a method for determining a push user group according to Embodiment 3 of the disclosure. This embodiment is an exemplary description of the above-mentioned embodiment. In this embodiment, the one-time push behavior is taken as an example for description. In an embodiment, referring to FIG. 4, the method includes:
  • Step 301 Determine whether it is the first push. If yes, go to step 302, otherwise, go to step 308.
  • whether it is the first push is determined by the number of message pushes.
  • Step 302 Obtain attribute tags of users in the first target user group.
  • the first target user group is a user group that has not accepted the push behavior.
  • the first target user group is limited to all users currently using the set application.
  • Step 303 Identify the attribute tag through the user classification model to determine the user level of the corresponding user.
  • Step 304 Calculate the estimated value of the first arrival rate of each user according to the user level and the estimated coefficient.
  • Step 305 Select a first percentage of users with the highest estimated first arrival rate from the first target user group to form a second target user group.
  • Step 306 Push messages to the second target user group.
  • Step 307 Record the message push data of each user.
  • Step 308 Determine whether the number of times of pushing the second message exceeds a set number of times. If it does not exceed the set number of times, step 309 is executed; otherwise, step 318 is executed.
  • the setting range is [1,10].
  • step 308 may also be to determine the relationship between the second message push times and the set times range. If the number of times of pushing the second message is less than the set number of times, return to step 302. If the second message push times are within the set times, step 309 is executed. If the number of times of pushing the second message is greater than the set number of times, step 318 is executed.
  • Step 309 Obtain attribute tags of users in the first target user group.
  • the first target user group is a user group who has not accepted the push behavior among all users who use the set application.
  • Step 310 Identify the attribute tag through the user classification model to determine the user level of the corresponding user.
  • Step 311 Calculate the estimated value of the first arrival rate of each user according to the user level and the estimated coefficient.
  • Step 312 Select the second percentage user with the highest estimated first arrival rate in the first target user group, randomly select the third percentage user from the remaining users in the first target user group, and select All of the users form the second target user group.
  • the third percentage is lower than the second percentage, and the second percentage is lower than the first percentage.
  • Step 313 Obtain user tags and message push data of users in the third target user group.
  • the third target user group refers to a user group that has accepted the push behavior among all users who use the set application.
  • Step 314 Obtain the second arrival rate estimation value of each user according to the user tag and corresponding message push data.
  • Step 315 Select a fourth percentage of users in the third target user group according to the second arrival rate estimate to form the fourth target user group.
  • Step 316 Push messages to the second target user group and the fourth target user group.
  • Step 317 Record the message push data of each user.
  • step 309-step 312 and step 313-step 315 can be executed simultaneously.
  • Step 318 Obtain attribute tags of users in the first target user group.
  • the first target user group is a user group who has not accepted the push behavior among all users who use the set application.
  • Step 319 Identify the attribute tag through the user classification model to determine the user level of the corresponding user.
  • Step 320 Calculate the estimated value of the first arrival rate of each user according to the user level and the estimated coefficient.
  • Step 321 Select a second percentage of users with the highest estimated first arrival rate from the first target user group to form a second target user group.
  • Step 322 Obtain user tags and message push data of users in the third target user group.
  • Step 323 Obtain the second arrival rate estimation value of each user according to the user tag and corresponding message push data.
  • Step 324 Select a fourth percentage of users in the third target user group according to the second estimated value of arrival rate to form the fourth target user group.
  • Step 325 Push messages to the second target user group and the fourth target user group.
  • Step 326 Record the message push data of each user.
  • the above process can ensure the rationality of the push user's choice every time a message is pushed.
  • FIG. 5 is a schematic structural diagram of an apparatus for determining a push user group provided in the fourth embodiment of the disclosure.
  • the device for determining a push user group includes: an acquisition module 401, a calculation module 402, a determination module 403, and a push module 404.
  • the obtaining module 401 is used to obtain the user level and estimated coefficient of each user in the first target user group.
  • the estimated coefficient is determined by the operation data of the corresponding user, and the operation data is the data generated when the user operates and sets the application;
  • the calculation module 402 is used to calculate the estimated value of the first arrival rate of each user according to the user level and the estimated coefficient;
  • the determination module 403 is used to select the first set in the first target user group according to the first estimated value of the arrival rate.
  • a certain number of users form the second target user group;
  • the push module 404 is used to push messages to the second target user group.
  • the first target user group is a user group that has not accepted the push behavior.
  • the acquisition module 401 includes: a tag acquisition unit for acquiring attribute tags of users in the first target user group; and a level determination unit for identifying attribute tags to determine the user level of the corresponding user.
  • the level determining unit is used to identify the attribute label through the user classification model to determine the user level of the corresponding user.
  • the user classification model is obtained by training the attribute label and user level of each user in the first reference user group.
  • a recording module which is used to record the message push data of each user in the second target user group after pushing the message to the second target user group.
  • the message push data includes: the first message push Number of times and/or number of message clicks.
  • a data acquisition module for acquiring user tags and message push data of users in the third target user group.
  • the user tags include: attribute tags, operation data, operation derived data, and user level At least one of the third target user group is the user group that has accepted the push behavior; the reach value determining module is used to obtain the second reach rate estimate value of each user according to the user tag and corresponding message push data; users The group selection module is used to select a second set number of users from the third target user group to form the fourth target user group according to the second estimated value of arrival rate; the message push module is used to send messages to the fourth target user group Push.
  • the arrival value determination module is used to: use the arrival rate estimation model to identify the user tag and the corresponding message push data to obtain the second arrival rate estimation value of each user, and the arrival rate estimation
  • the model is obtained by training the user tags of each user in the second reference user group, the message push data and the estimated value of the arrival rate.
  • the determining module 403 includes: a frequency determining unit for determining the second message push times of the set application; a user group determining unit for predicting the second message push times and the first arrival rate. Estimate, select a first set number of users from the first target user group to form the second target user group.
  • the user group determining unit includes: a first subunit, used for when the number of pushes of the second message is within a set number of times, according to the order of the first arrival rate estimation value from high to low, Select the first number of users in the first target user group to form the first subgroup, and randomly select the second number of users from the remaining users to form the second subgroup, and the first subgroup and the second subgroup form the second target User group; the second subunit is used to select the third number of the first target user group in the first target user group in the order of the first arrival rate estimate value when the second message push times are outside the set number range Users form the second target user group.
  • the push module 404 includes: a push method determining unit for determining the current push method, the current push method includes: pop-up push and/or information push; a push message unit for performing according to the current push method News push.
  • the current push method includes pop-up push
  • the arrival rate prediction model includes the pop-up arrival rate prediction model
  • the current push method includes information push
  • the arrival rate estimation model includes the information arrival rate estimation model
  • the device for determining the push user group provided by the embodiments of the present disclosure is included in the device for determining the push user group, and can be used to execute the method for determining the push user group provided by any of the foregoing embodiments, and has corresponding functions and beneficial effects.
  • FIG. 6 is a schematic structural diagram of a device for determining a push user group provided by Embodiment 5 of the disclosure.
  • the determining push user group includes a processor 50, a memory 51, an input device 52, an output device 53, and a communication device 54; the number of processors 50 in the device for determining push user group may be one or more ,
  • Figure 6 takes a processor 50 as an example; the processor 50, memory 51, input device 52, output device 53, and communication device 54 in the device that determines the push of the user group can be connected by a bus or other means, as shown in Figure 6 Take the bus connection as an example.
  • the memory 51 can be used to store software programs, computer-executable programs, and modules, such as the program instructions/modules corresponding to the method for determining the push user group in the embodiment of the present disclosure (for example, determine the push user group).
  • the processor 50 runs the software programs, instructions, and modules stored in the memory 51 to execute various functional applications and data processing of the device that determines the push user group, that is, realizes the aforementioned method for determining the push user group.
  • the memory 51 may mainly include a storage program area and a storage data area.
  • the storage program area may store an operating system and an application program required by at least one function; the storage data area may store data created according to the use of the device that determines the push user group Wait.
  • the memory 51 may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one magnetic disk storage device, a flash memory device, or other non-volatile solid-state storage devices.
  • the memory 51 may include a memory remotely provided with respect to the processor 50, and these remote memories may be connected to a device that determines the user group to be pushed through a network. Examples of the aforementioned networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.
  • the input device 52 may be used to receive inputted numeric or character information, and generate key signal input related to determining user settings and function control of the device for pushing the user group.
  • the output device 53 may include a display device such as a display screen.
  • the communication device 54 is used for data communication with a device corresponding to the setting application.
  • the aforementioned device for determining a push user group includes a device for determining a push user group, which can be used to execute any method for determining a push user group, and has corresponding functions and beneficial effects.
  • Embodiments of the present disclosure also provide a storage medium containing computer-executable instructions, when the computer-executable instructions are executed by a computer processor, a method for determining a push user group is executed, and the method includes:
  • the estimated coefficient is determined by the operation data of the corresponding user, and the operation data is the data generated when the user operates the application program;
  • a storage medium containing computer-executable instructions provided by an embodiment of the present disclosure is not limited to the method operations described above, and can also perform the push user group determination provided by any embodiment of the present disclosure. Related operations in the method.
  • the present disclosure can be implemented by software and necessary general-purpose hardware, of course, it can also be implemented by hardware, but in many cases the former is a more suitable implementation .
  • the technical solution of the present disclosure can be embodied in the form of a software product, which can be stored in a computer-readable storage medium, such as a computer floppy disk, Read-Only Memory (ROM), Random Access Memory (RAM), Flash memory (FLASH), hard disk or optical disk, etc., including several instructions to make a computer device (which can be a personal computer, A server, or a network device, etc.) execute the method described in each embodiment of the present disclosure.
  • the units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding function can be realized;
  • the specific names of the functional units are only used to facilitate distinguishing from each other, and are not used to limit the protection scope of the present disclosure.

Landscapes

  • Business, Economics & Management (AREA)
  • Strategic Management (AREA)
  • Engineering & Computer Science (AREA)
  • Accounting & Taxation (AREA)
  • Development Economics (AREA)
  • Finance (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Game Theory and Decision Science (AREA)
  • Economics (AREA)
  • Marketing (AREA)
  • Physics & Mathematics (AREA)
  • General Business, Economics & Management (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Information Transfer Between Computers (AREA)

Abstract

La présente invention concerne le domaine technique de l'envoi d'informations, et concerne un procédé, appareil, et un dispositif de détermination d'un groupe d'utilisateurs d'envoi, et un support d'informations, comprenant : l'obtention d'un niveau d'utilisateur et d'un coefficient d'estimation de chaque utilisateur dans un premier groupe d'utilisateurs cibles, le coefficient d'estimation étant déterminé au moyen de données de fonctionnement de l'utilisateur correspondant, et les données de fonctionnement étant des données générées lorsque l'utilisateur fait fonctionner une application définie; le calcul d'une première valeur d'estimation de taux d'arrivée de chaque utilisateur en fonction du niveau d'utilisateur et du coefficient d'estimation; la sélection d'un premier nombre défini d'utilisateurs à partir du premier groupe d'utilisateurs cibles en fonction de la première valeur d'estimation de taux d'arrivée pour former un second groupe d'utilisateurs cibles; et l'envoi d'informations vers le second groupe d'utilisateurs cibles. L'utilisation du procédé peut résoudre le problème dans l'état de la technique associé de manque de rationalité dans le mode de détermination de cible d'envoi.
PCT/CN2019/126717 2019-06-25 2019-12-19 Procédé, appareil et dispositif de détermination d'un groupe d'utilisateurs d'envoi et support de stockage WO2020258773A1 (fr)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201910556881.8A CN110264276A (zh) 2019-06-25 2019-06-25 确定推送用户群的方法、装置、设备及存储介质
CN201910556881.8 2019-06-25

Publications (1)

Publication Number Publication Date
WO2020258773A1 true WO2020258773A1 (fr) 2020-12-30

Family

ID=67921436

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2019/126717 WO2020258773A1 (fr) 2019-06-25 2019-12-19 Procédé, appareil et dispositif de détermination d'un groupe d'utilisateurs d'envoi et support de stockage

Country Status (2)

Country Link
CN (1) CN110264276A (fr)
WO (1) WO2020258773A1 (fr)

Families Citing this family (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110264276A (zh) * 2019-06-25 2019-09-20 广州视源电子科技股份有限公司 确定推送用户群的方法、装置、设备及存储介质
CN115023933A (zh) * 2020-02-28 2022-09-06 深圳市欢太科技有限公司 内容推送方法、装置、服务器及存储介质
CN112801685B (zh) * 2020-09-10 2024-06-18 腾讯科技（深圳）有限公司 信息推送方法、装置、计算机设备及存储介质
CN112653769A (zh) * 2021-01-08 2021-04-13 青岛海尔科技有限公司 一种消息的推送方法及系统

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101547162A (zh) * 2008-03-28 2009-09-30 国际商业机器公司 基于用户的状态信息标签用户的方法及装置
CN101997894A (zh) * 2009-08-14 2011-03-30 阿里巴巴集团控股有限公司 一种信息推送方法及其系统和网络系统
CN106294778A (zh) * 2016-08-11 2017-01-04 北京小米移动软件有限公司 信息推送方法和装置
CN107360246A (zh) * 2017-07-28 2017-11-17 广州优视网络科技有限公司 一种消息推送方法及装置、一种终端及存储介质
CN108764994A (zh) * 2018-05-24 2018-11-06 深圳前海桔子信息技术有限公司 一种用户行为指引方法、装置、服务器和存储介质
CN110264276A (zh) * 2019-06-25 2019-09-20 广州视源电子科技股份有限公司 确定推送用户群的方法、装置、设备及存储介质

Family Cites Families (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103729359B (zh) * 2012-10-12 2017-03-01 阿里巴巴集团控股有限公司 一种推荐搜索词的方法及系统
CN105045831B (zh) * 2015-06-30 2018-04-13 北京奇艺世纪科技有限公司 一种消息推送方法及装置
CN106251174A (zh) * 2016-07-26 2016-12-21 北京小米移动软件有限公司 信息推荐方法及装置
CN106557956A (zh) * 2016-11-29 2017-04-05 国网山东省电力公司电力科学研究院 一种关于配置客户缴费服务信息推送策略的方法
US11107025B2 (en) * 2016-12-13 2021-08-31 STREAM METHODS, Inc. System and method for producing and distributing information relevant to water events
CN107613022B (zh) * 2017-10-20 2020-10-16 阿里巴巴（中国）有限公司 内容推送方法、装置及计算机设备

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101547162A (zh) * 2008-03-28 2009-09-30 国际商业机器公司 基于用户的状态信息标签用户的方法及装置
CN101997894A (zh) * 2009-08-14 2011-03-30 阿里巴巴集团控股有限公司 一种信息推送方法及其系统和网络系统
CN106294778A (zh) * 2016-08-11 2017-01-04 北京小米移动软件有限公司 信息推送方法和装置
CN107360246A (zh) * 2017-07-28 2017-11-17 广州优视网络科技有限公司 一种消息推送方法及装置、一种终端及存储介质
CN108764994A (zh) * 2018-05-24 2018-11-06 深圳前海桔子信息技术有限公司 一种用户行为指引方法、装置、服务器和存储介质
CN110264276A (zh) * 2019-06-25 2019-09-20 广州视源电子科技股份有限公司 确定推送用户群的方法、装置、设备及存储介质

Also Published As

Publication number Publication date
CN110264276A (zh) 2019-09-20

Similar Documents

Publication Publication Date Title
WO2020258773A1 (fr) Procédé, appareil et dispositif de détermination d'un groupe d'utilisateurs d'envoi et support de stockage
CN110458220B (zh) 人群定向方法、装置、服务器及存储介质
US20160307131A1 (en) Method, apparatus, and system for controlling delivery task in social networking platform
US20210248198A1 (en) Content Recommendation Method and Apparatus, Mobile Terminal, and Server
CN114265979B (zh) 确定融合参数的方法、信息推荐方法和模型训练方法
US20160132904A1 (en) Influence score of a brand
CN108959319B (zh) 信息推送方法和装置
CN111405030B (zh) 一种消息推送方法、装置、电子设备和存储介质
WO2016107354A1 (fr) Procédé et appareil permettant la mise en œuvre d'une poussée de messages de ressources personnalisés en fonction de l'utilisateur
CN111523035B (zh) App浏览内容的推荐方法、装置、服务器和介质
CN112241327A (zh) 分享信息处理方法、装置、存储介质与电子设备
CN106777354A (zh) 推广信息新鲜度确定方法及装置
WO2021081914A1 (fr) Procédé et appareil de détermination d'objet à pousser, dispositif terminal et support de stockage
CN114461919A (zh) 信息推荐的模型训练方法及装置
CN112269918B (zh) 信息推荐方法、装置、设备及存储介质
CN112287208B (zh) 用户画像生成方法、装置、电子设备及存储介质
CN108388652B (zh) 一种发送歌单标识的方法和装置
CN115329078B (zh) 文本数据处理方法、装置、设备以及存储介质
CN114258670A (zh) 信息推送方法、装置、电子设备以及存储介质
CN115809889A (zh) 基于营销效果的智能客群筛选方法、系统、介质及设备
WO2018161306A1 (fr) Recommandation d'application
CN111275473B (zh) 内容项投放方法、装置、服务器及存储介质
CN113269600B (zh) 一种信息发送的方法和装置
CN113538030B (zh) 一种内容推送方法、装置及计算机存储介质
US12130799B2 (en) Data integrity optimization

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 19934428

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 19934428

Country of ref document: EP

Kind code of ref document: A1