CN113590927A

CN113590927A - Content recommendation method, device, server and storage medium

Info

Publication number: CN113590927A
Application number: CN202010366974.7A
Authority: CN
Inventors: 谢楠
Original assignee: Lumi United Technology Co Ltd
Current assignee: Shenzhen Lumi United Technology Co Ltd; Lumi United Technology Co Ltd
Priority date: 2020-04-30
Filing date: 2020-04-30
Publication date: 2021-11-02

Abstract

The embodiment of the application discloses a content recommendation method, a content recommendation device, a server and a storage medium. The method comprises the following steps: the method comprises the steps of acquiring user data in a target application program collected based on buried point information, wherein the user data comprise operation data and/or a user source of the target application program used by a user at this time, acquiring exit probability according to the operation data and/or the user source, namely predicting the probability that the user exits the target application program according to the operation data and/or the user source of the user in real time, and pushing content according to the exit probability and the operation data after the exit probability is acquired. The method comprises the steps of acquiring the probability of the user exiting the target application program through operation data and/or a user source in real time, pushing content related to the operation of the user when the user still uses the target application program by combining the operation data of the user and the exiting probability, promoting the effectiveness of content recommendation, saving the user in time and further promoting the viscosity of the user.

Description

Content recommendation method, device, server and storage medium

Technical Field

The present application relates to the field of computer technologies, and in particular, to a content recommendation method, apparatus, server, and storage medium.

Background

With the development of science and technology, electronic devices such as mobile terminals, computers or tablets and the like are rapidly developed, generally, various application programs can be installed on the electronic devices, wherein different application programs can realize different functions, and people can meet various requirements in life or work through the application programs.

However, the applications on the market are of a wide variety and users have a variety of choices, so that the user of the applications is not very sticky.

Disclosure of Invention

The embodiment of the application provides a content recommendation method, a content recommendation device, a server and a storage medium, so as to solve the problems.

In a first aspect, an embodiment of the present application provides a content recommendation method, where the method includes: acquiring user data in a target application program collected based on the embedded point information, wherein the user data comprises operation data and/or a user source of the target application program used by a user at this time; obtaining exit probability according to the operation data of the user and/or a user source, wherein the exit probability is the probability that the user corresponding to the user data exits the target application program; and pushing content according to the exit probability and/or the operation data.

In a second aspect, an embodiment of the present application provides a content recommendation apparatus, including: the system comprises a first acquisition module, a second acquisition module and a third acquisition module, wherein the first acquisition module is used for acquiring user data in a target application program collected based on embedded point information, and the user data comprises operation data and/or a user source of the target application program used by a user at this time; a second obtaining module, configured to obtain an exit probability according to the operation data of the user and/or a user source, where the exit probability is a probability that a user corresponding to the user data exits the target application program; and the pushing module is used for pushing the content according to the exit probability and the operation data.

In a third aspect, embodiments of the present application provide a server including one or more processors, a memory, and a computer program stored on the memory and executable on the processors, where the computer program, when executed by the processors, implements the method applied to the server as described above.

In a fourth aspect, embodiments of the present application provide a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method described above.

According to the content recommendation method, the content recommendation device, the server and the storage medium, user data in a target application program collected based on the embedded point information are obtained, the user data comprise operation data and/or user sources of the target application program used by a user at this time, exit probability is obtained according to the operation data and/or the user sources, namely the probability that the user exits the target application program can be predicted according to the operation data and/or the user sources of the user in real time, and after the exit probability is obtained, content is pushed according to the exit probability and/or the operation data. The method comprises the steps of acquiring the probability of the user exiting the target application program through operation data and/or a user source in real time, pushing content related to the operation of the user when the user still uses the target application program by combining the operation data of the user and the exiting probability, promoting the effectiveness of content recommendation, saving the user in time and further promoting the viscosity of the user.

These and other aspects of the present application will be more readily apparent from the following description of the embodiments.

Drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed to be used in the description of the embodiments are briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present application, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts.

Fig. 1 shows a schematic application environment applicable to the content recommendation method provided in the embodiment of the present application.

Fig. 2 shows a flowchart of a content recommendation method according to an embodiment of the present application.

Fig. 3 shows a flowchart of a content recommendation method according to another embodiment of the present application.

Fig. 4 shows a flow chart of part of the steps of a content recommendation method provided on the basis of the embodiment provided in fig. 3.

Fig. 5 is a flowchart illustrating a content recommendation method according to still another embodiment of the present application.

FIG. 6 is a flowchart illustrating a content recommendation method according to still another embodiment of the present application

Fig. 7 shows a flow chart of part of the steps of a content recommendation method provided on the basis of the embodiment provided in fig. 6.

Fig. 8 is a functional block diagram of a content recommendation apparatus according to an embodiment of the present application.

Fig. 9 is a block diagram illustrating a structure of a server for executing a content recommendation method according to an embodiment of the present application.

Detailed Description

In order to make the technical solutions better understood by those skilled in the art, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application.

As for the application programs, because of the wide variety of application programs, competition among the developers of the application programs is increasingly intense. User stickiness refers to the loyalty of a user to a brand or product, and in the present embodiment, may be understood as the loyalty of an application. For example, a user's mobile terminal is installed with shopping software a and software B, the user often purchases an item in software a, and then the user's viscosity of software a is greater than that of software B for the user. It can be understood that, the more the user is sticky, the more the number of people using the application, the higher the click volume or the volume of the transaction in the application, and thus the application can occupy a certain market status, and therefore, the improvement of the user's stickiness is a problem to be considered by developers of each application.

Most application developers push the push content to the user after the user exits the application in order to increase the user's viscosity, thereby attracting the user to click on the push content and reuse the application. However, since the user already exits the application program, the method of recalling the user by pushing the content is not effective. The inventor finds in research that if the quitting intention of the user can be accurately predicted when the user uses the application program, the content can be pushed in time when the quitting intention of the user is predicted, the probability of saving the user can be increased, and therefore the viscosity of the user is improved. If the user's intention is predicted by using the data of registration time, gender, occupation, etc., it is difficult to accurately predict the user's intention to quit.

Therefore, the inventor proposes a content recommendation method according to an embodiment of the present application, which obtains user data in a target application program collected based on embedded point information, where the user data includes operation data and/or a user source of the target application program used by a user this time, and obtains an exit probability according to the operation data and/or the user source, that is, a probability that the user exits the target application program can be predicted in real time according to the operation data of the user and/or the user source, and after obtaining the exit probability, content is pushed according to the exit probability and the operation data. The method comprises the steps of acquiring the probability of the user exiting the target application program through operation data and/or a user source in real time, pushing content related to the operation of the user when the user still uses the target application program by combining the operation data of the user and the exiting probability, promoting the effectiveness of content recommendation, saving the user in time and further promoting the viscosity of the user. .

The embodiments of the present application will be described in detail below with reference to the accompanying drawings.

Referring to fig. 1, fig. 1 illustrates an application environment 10 of a content recommendation method according to an embodiment of the present application. The application environment 10 includes: a local/cloud server 11, an electronic device 12, and a network 13. The local server/cloud server 11 performs data interaction with the electronic device 12 through a network 13, where the network 13 may be a wide area network or a local area network, or a combination of the two, and uses a wireless link to implement data transmission. The electronic device 12 may be a mobile phone, a tablet computer, a pc (personal computer), a notebook computer, a smart television, a vehicle-mounted terminal, or the like.

The local server/cloud server 11 and the electronic device 12 are connected through a network 13 to implement content recommendation. In this embodiment of the present application, the training of the prediction model may be performed in the local server/cloud server 11, and after the prediction model is trained, the local server/cloud server 11 may receive user data including operation data and/or user source data of the target application program used by the user this time and sent by the electronic device 12, input the user data into the trained prediction model to obtain an exit probability, and send the push content to the electronic device 11 through the network 13 for display according to the exit probability and the operation data push content.

There are other embodiments for training the predictive model and enabling content push based on the application environment.

In some embodiments, the training of the prediction model may be performed in the electronic device 12, and after the prediction model is trained, the obtained operation data including the current use of the target application by the user and/or the user data from the user source is input into the trained prediction model to obtain an exit probability, so that the pushed content is output and displayed on the electronic device 12 according to the exit probability and the push content of the operation data. The electronic device 12 may also send the trained parameters of the prediction model to the local server/cloud server 11 through the network 13 for backup.

In some embodiments, the prediction model may be trained in the local server/cloud server 11, and the trained prediction model may be transmitted to the electronic device 11 through the network 13. The electronic device 11 may input the operation data including the current use of the target application by the user and/or the user data from the user source into the trained prediction model to obtain the exit probability, so as to display the pushed content on the electronic device 12 according to the exit probability and the pushed content of the operation data.

In the embodiment of the application, the prediction model is mainly trained in the local server/cloud server 11, so that the trained prediction model can be used for acquiring the exit probability of the user in real time, content push is performed in time, and the viscosity of the user is improved.

Referring to fig. 2, an embodiment of the present application provides a content recommendation method, which may be applied to a server, and the method may include:

step S110: and acquiring user data in the target application program collected based on the buried point information, wherein the user data comprises operation data and/or user sources of the target application program used by the user at this time.

The buried point refers to the related technology and implementation process thereof for capturing, processing and transmitting specific user behaviors or events. E.g., the number of clicks of the user, the length of time a certain video is viewed, etc. The counting of the buried points is essentially to monitor events in the running process of the software application, and judge and capture when the events needing attention occur. The point burying technology can be divided into code point burying, visual point burying and non-buried point burying. The code embedding point is formed by adding some codes in a webpage/app, data reporting is carried out when a user triggers corresponding behaviors, and the code embedding point can set a certain event custom attribute in detail, but the workload of the embedding point is large. The visualization embedded point is characterized in that events can be configured through a visualization interface by using a visualization interaction means, data are collected, and the visualization embedded point can only support client behaviors without intervention of developers. The non-embedded point means that after a developer collects the SDK in an integrated mode, the SDK directly starts to capture and detect all behaviors of the user in the application and reports all the behaviors, extra codes do not need to be added by the developer, and the data reporting is supported first and then the embedded point is carried out.

Different embedding point modes correspond to different advantages and disadvantages, and a proper embedding point mode can be selected by combining with a specific use scene, and the specific embedding point mode is not limited herein.

And defining the target application program as the application program used by the user, wherein the target application program can be pre-buried with points, and the user data in the target application program can be collected based on the buried point information. The user data includes operation data of the target application program used by the user this time, that is, when the user uses the target application program, all operation data corresponding to the embedded point information may be collected, for example, behaviors such as clicking, browsing, placing an order, and the like, and may also include time information of the behaviors, that is, the operation data may include operation steps and a time sequence between the operation steps, that is, a sequence between the operation steps. Optionally, the time sequence of the operation steps may further include a time interval between the operation steps. The user data may also include a user source, and for the target application, there are various ways how the user enters the target application, for example, the user may start the target application through a web page of a browser, or jump to the target application through a link in another application, and start the target application from a different source, which also affects the probability that the user exits the target application. For example, if the user opens the target application program from the installed application programs, the user actively uses the target application program, and the intention of exiting the target application program is not strong. Therefore, the source of the user can be buried, and the source of the target application program used by the user at this time can be obtained.

After the operation data and/or the user source of the user are obtained, the next step may be executed according to the obtained operation data and/or the user source of the user.

Step S120: and obtaining exit probability according to the operation data of the user and/or a user source, wherein the exit probability is the probability that the user corresponding to the user data exits the target application program.

After the operation data and/or the user source of the user are acquired, an exit probability can be acquired according to the acquired operation data and/or the user source of the user, wherein the exit probability is the probability that the user corresponding to the user data exits the target application program.

When the user uses the target application, various operations may be performed, for example, the target application used by the user is shopping software a, and when the user uses the shopping software, the user may perform searching, browsing, and the like to check items that the user wants to purchase, however, after browsing, the user may directly exit the shopping software a and not make a purchase. Then the exit probability is the probability that the user exits the target application. Wherein a greater value of the exit probability indicates a greater likelihood that the user will exit the target application. For example, a 20% exit probability indicates that the user's intent to exit the target application is not strong, and a 100% exit probability indicates that the user's intent to exit the target application is strong.

The obtaining of the exit probability according to the operation data of the user and/or the user source may depend on a deep learning model, or may be a pre-trained prediction model, and the prediction model is used for outputting the exit probability according to the input operation data and/or the user source, so that the obtained operation data of the user and/or the user source is input into the trained prediction model to obtain the corresponding exit probability.

Step S130: and pushing content according to the exit probability and the operation data.

After the exit probability is obtained, the exit probability and the exit threshold value can be compared, the content is pushed according to the operation data, the user is attracted to continue using the target application program through the pushed content, and the user viscosity of the application program is improved.

In some embodiments, when the exit probability is obtained, the operation information before the user, which is obtained according to the operation data, includes browsing content, search content, and the like, so that the content which the user is interested in can be pushed in combination with the operation data of the user, the attraction of the pushed content to the user is improved, and the user viscosity of the application program is improved.

In some embodiments, when the exit probability is obtained, an exit threshold range in which the exit probability is located may be determined, a category of the pushed content is determined according to the exit threshold range, and the content related to the category is pushed.

According to the content recommendation method provided by the embodiment of the application, the user data in the target application program collected based on the embedded point information is obtained, the user data comprises the operation data and/or the user source of the target application program used by the user at this time, the quitting probability is obtained according to the operation data and/or the user source, the probability that the user quits the target application program can be predicted according to the operation data and/or the user source of the user in real time, and after the quitting probability is obtained, the content is pushed according to the quitting probability and the operation data. The method comprises the steps of acquiring the probability of the user exiting the target application program through operation data and/or a user source in real time, pushing content related to the operation of the user when the user still uses the target application program by combining the operation data and/or the exiting probability of the user, promoting the effectiveness of content recommendation, saving the user in time and further promoting the viscosity of the user. .

Referring to fig. 3, another embodiment of the present application provides a content recommendation method, which focuses on describing a process of obtaining an exit probability according to operation data of a user and/or a user source based on the above embodiment, and the method may include:

step S210: and acquiring user data in the target application program collected based on the buried point information, wherein the user data comprises operation data and/or user sources of the target application program used by the user at this time.

Step S210 may refer to corresponding parts of the foregoing embodiments, and will not be described herein again.

Step S220: and inputting the operation data and/or the user source into a pre-trained prediction model to obtain the exit probability.

Inputting the acquired operation data and/or user sources into a pre-trained prediction model, and acquiring exit probabilities corresponding to the operation data and/or the user sources, wherein the operation data can comprise operation steps and time sequences among the operation steps. The same operation steps, but different timing sequences between the operation steps may also result in different exit probabilities.

The prediction model may be a neural network, the neural network includes at least one input layer, a hidden layer and an output layer, and the corresponding activation function is selected to operate the operation data and/or the user source input to the neural network to output the corresponding exit probability, and the neural network used by the prediction model may be a Long-Short Term Memory artificial neural network (LSTM).

The prediction model used here is a trained model, the input data of the model is the operation data and/or user source of the user, and the output data is the exit probability, that is, the probability that the user corresponding to the operation data and/or user source exits the target application program.

It will be appreciated that the predictive model may need to be trained before the exit probability is obtained using the predictive model. Specifically, referring to fig. 4, the process of training the prediction model specifically includes the following steps:

step S221: acquiring a plurality of groups of different sample data and a preset exit probability corresponding to each group of sample data, wherein the sample data comprises operation data and/or a user source.

Before training the prediction model, a plurality of groups of different sample data need to be acquired, and the exit probability corresponding to the sample data can be obtained by inputting the sample data into the prediction model. When sample data is acquired, a preset exit probability corresponding to the sample data, namely a real exit probability corresponding to the sample data, can be acquired.

The source of the multiple sets of sample data may be operation data and/or a user source for obtaining the use of the target application program by the user in the historical period, wherein the operation data generated between the opening of the target application program and the exiting of the target application program and the source for opening the target application program at the time serve as one set of sample data. The historical period refers to a certain period in the past, and may be, for example, 30 days in the past, a week in the past, and the like, and specifically, the historical period may be set according to actual needs, and is not limited herein. For example, the historical period is one week, and within the one week, the user uses the target application program 3 times in total, then 3 sets of sample data may be obtained, and the prediction model may be trained by using three sets of sample data.

Step S222: and inputting each group of the sample data into the prediction model to obtain corresponding exit probability.

And inputting the operation data and/or the user source into the prediction model to obtain the exit probability of the sample data corresponding to the group.

Step S223: and when the difference value between the exit probability corresponding to the multiple groups of sample data and the preset exit probability is smaller than the preset value, obtaining a trained prediction model.

When each group of sample data is input into the prediction model, the exit probability corresponding to the sample data can be obtained, and each group of sample data corresponds to one preset exit probability, so that the parameters of the prediction model can be adjusted according to the exit probability and the preset exit probability, and when the difference value between the exit probability obtained by inputting each group of sample data into the prediction model and the preset exit probability corresponding to the sample data is smaller than the preset value, the prediction model can be considered to be trained.

Step S230: and pushing content according to the exit probability and the operation data.

Step S230 may refer to corresponding parts of the foregoing embodiments, and will not be described herein again.

The content recommendation method provided by the embodiment of the application acquires user data in a target application program collected based on embedded point information, wherein the user data comprises operation data and/or a user source of the target application program used by a user at this time; inputting the operation data and/or the user source into a pre-trained prediction model to obtain the exit probability; and pushing content according to the exit probability and the operation data. When the preset model is trained or used, the data of the input prediction model are operation data and/or user sources, wherein the operation data comprise operation steps and time sequences among the operation steps, so that the quitting probability can be obtained more accurately, more accurate push content can be carried out according to the quitting probability, and the viscosity of a user is improved.

Referring to fig. 5, a further embodiment of the present application provides a content recommendation method, which focuses on describing a process of pushing content according to the exit probability and the operation data on the basis of the previous embodiment, and the method may include:

step S310: and acquiring user data in the target application program collected based on the buried point information, wherein the user data comprises operation data and/or user sources of the target application program used by the user at this time.

Step S320: and obtaining exit probability according to the operation data of the user and/or a user source, wherein the exit probability is the probability that the user corresponding to the user data exits the target application program.

The steps S310 to S320 refer to corresponding parts of the foregoing embodiments, and are not described herein again.

Step S330: judging whether the exit probability is greater than or equal to an exit threshold value; if yes, go to step S340; if not, go to step S320.

When the exit probability is obtained, the exit probability is compared with the exit threshold, whether the exit probability is greater than or equal to the exit threshold is determined, if the exit probability is greater than or equal to the exit threshold, it indicates that the intention of the user to exit the target application is strong, and the target application is likely to exit in the next operation, and step S340 may be executed. If the exit probability is less than the exit threshold, it indicates that the intention of the user to exit the target application is not strong, and the possibility of exiting the target application in the next operation is not high, step S320 may be continuously performed to continuously obtain the exit probability according to the collected operation data and/or the user source.

Wherein the exit threshold is pre-stored. In some embodiments, the exit threshold may be derived using a trained predictive model. Specifically, a plurality of different sets of operation data and/or user sources may be obtained, where the plurality of different sets of operation data and/or user sources may be different sets of operation data, different sets of user sources, or different sets of operation data and different sets of user sources. The different operation data have different operation steps or different time sequences among the operation steps. For example, the operation data is different operation data in that the search box is clicked first and then the recommended content is browsed, and the operation data is different from the operation data in that the recommended content is browsed first and then the search box is clicked.

And inputting the multiple groups of different time sequence operation data and/or user sources into the trained prediction model to obtain a plurality of corresponding exit probabilities. A probability value that is most representative of the user exiting the application is determined from the plurality of exit probabilities as a probability threshold, which may be, for example, the maximum of the plurality of exit probabilities.

Step S340: pushing the content related to the last operation step in the time sequence.

When the exit probability is determined to be greater than or equal to the exit threshold, indicating that the user has a greater likelihood of exiting the target application in the next operation, the content may be pushed at this point to attract the user in order to timely withhold the user.

The user data acquired based on the buried point information comprises operation data and/or user sources, wherein the operation data comprises operation steps and time sequences among the operation steps. Then, when the exit probability is greater than or equal to the exit threshold, the content may be pushed, and a chronological last operation step may be acquired, and the content related to the operation step may be pushed to attract the user.

For example, the target application program is shopping software, the operation data of the user acquired at the moment is click search box-search washing machine-browse xx-brand washing machine 1s, wherein the appearance sequence represents the time sequence of the operation steps, namely, the search box is clicked first, then the washing machine is searched, then the xx-brand washing machine 1s is browsed, and the user source is a WeChat link. If the quit probability obtained according to the operation data and/or the user source is 90% and the quit threshold is 80%, if the quit probability is greater than the quit threshold, the last operation step in the time sequence can be obtained by browsing the xx-brand washing machine, and the content related to the xx-brand washing machine can be pushed, such as pushing a coupon of the washing machine, and an obtaining button of the coupon is displayed on a display page of the washing machine. So as to attract the user to click the coupon acquisition button, check the coupon or continue browsing the goods, and the like, thereby saving the user. If the target application is a content-type application, such as news software, the last news browsed by the user may be obtained, and the content related to the news may be pushed. In some embodiments, the content related to the news with the longest browsing time can be pushed.

The content recommendation method provided by the embodiment of the application acquires user data in a target application program collected based on embedded point information, wherein the user data comprises operation data and/or a user source of the target application program used by a user at this time, and acquires exit probability according to the operation data and/or the user source; judging whether the exit probability is greater than or equal to an exit threshold value; if yes, pushing the content related to the last operation step in the time sequence. The method comprises the steps of obtaining the probability of the user exiting the target application program through operation data and/or a user source, pushing content related to the operation of the user when the user still uses the target application program by combining the operation data of the user and the exiting probability, promoting the effectiveness of content recommendation, saving the user in time and further promoting the viscosity of the user. .

Referring to fig. 6, a further embodiment of the present application provides a content recommendation method, where on the basis of the previous embodiment, a process of pushing content according to the exit probability and the operation data is described in detail, and the method may include:

step S410: and acquiring user data in the target application program collected based on the buried point information, wherein the user data comprises operation data and/or user sources of the target application program used by the user at this time.

Step S420: and obtaining exit probability according to the operation data of the user and/or a user source, wherein the exit probability is the probability that the user corresponding to the user data exits the target application program.

Steps S410 to S420 can refer to the corresponding steps of the previous embodiments, and are not described again.

Step S430: determining an exit threshold range in which the exit probability is located.

Step S440: and acquiring the push content according to the exit threshold range of the exit probability.

The exit threshold range is prestored, when the exit probability is obtained, the exit threshold range where the exit probability is located can be determined, and the push content is obtained according to the exit threshold range where the exit probability is located.

Wherein the exit threshold range is pre-stored. In some embodiments, the exit threshold range may be derived using a trained predictive model. Specifically, a plurality of sets of different operation data and/or user sources may be obtained, where operation steps in the different operation data are different, or timing between operations is different. For example, the operation data is different operation data in that the search box is clicked first and then the recommended content is browsed, and the operation data is different from the operation data in that the recommended content is browsed first and then the search box is clicked.

And inputting the multiple groups of different time sequence operation data and/or user sources into the trained prediction model to obtain a plurality of corresponding exit probabilities. And dividing an exit threshold range according to the exit probabilities. Therefore, the push content can be obtained according to the exit threshold range where the exit probability is located, specifically, referring to fig. 7, a specific step of obtaining the push content according to the exit threshold range where the exit probability is located is shown:

step S441: and searching a corresponding relation table, and determining the category of the push content corresponding to the exit threshold range in which the exit probability is located.

A corresponding relation table is stored in advance, the corresponding relation table includes a corresponding relation between the exit threshold range and the category of the push content, and the specific corresponding relation may be preset by a developer. Specifically, the content of the correspondence table can refer to table 1.

TABLE 1

Out of threshold range	a～b	b～c	c～d
				Categories	Class 1	Class 2	Class 3

In table 1, if the exit probability is greater than or equal to a and less than b, the category of the corresponding push content is category 1; if the exit probability is greater than or equal to b and less than c, the corresponding type of the push content is type 2; if the exit probability is greater than or equal to c and less than d, the category of the corresponding push content is category 3.

Step S442: pushing the push content associated with the category.

When the category of the pushed content is determined through the corresponding relation table, the content related to the category can be directly pushed. For example, the target application program is shopping software, the quit probability is 70%, the quit threshold range is 0-40%, the corresponding category is promotion activity, the corresponding category is coupon when the quit threshold range is 40-80%, and the corresponding category is coupon when the quit threshold range is 80-100%.

At this time, the quit probability is 70%, the quit threshold range is 40% -80%, the corresponding category is the coupon, and the coupon of various commodities can be obtained and pushed.

The content recommendation method provided by the embodiment of the application acquires user data in a target application program collected based on embedded point information, wherein the user data comprises operation data and/or a user source of the target application program used by a user at this time, and acquires exit probability according to the operation data and/or the user source; determining an exit threshold range in which the exit probability is located; and acquiring the push content according to the threshold range where the exit probability is located. The method comprises the steps of obtaining the probability of the user exiting the target application program through operation data and/or a user source, pushing content related to the operation of the user when the user still uses the target application program by combining the operation data of the user and the exiting probability, promoting the effectiveness of content recommendation, saving the user in time and further promoting the viscosity of the user. .

Referring to fig. 8, which illustrates a content recommendation apparatus 500 provided in an embodiment of the present application and applicable to a server, the content recommendation apparatus 500 includes a first obtaining module 510, a second obtaining module 520, and a pushing module 530. The first obtaining module 510 is configured to obtain user data in a target application program collected based on the embedded point information, where the user data includes operation data and/or a user source of the target application program used by a user this time; the second obtaining module 520 is configured to obtain an exit probability according to the operation data of the user and/or a user source, where the exit probability is a probability that a user corresponding to the user data exits the target application program; the pushing module 530 is configured to push content according to the exit probability and the operation data.

The method comprises the steps of acquiring the probability of the user exiting the target application program through operation data and/or a user source in real time, pushing content related to the operation of the user when the user still uses the target application program by combining the operation data of the user and the exiting probability, promoting the effectiveness of content recommendation, saving the user in time and further promoting the viscosity of the user. .

Further, the second obtaining module 520 is further configured to input the operation data and/or the user source into a pre-trained prediction model to obtain the exit probability, where the prediction model is configured to output a corresponding exit probability according to the input operation data and/or the user source.

Further, the content recommendation device 500 further includes a training module, before inputting the operation data and/or the user source of the user into a pre-trained prediction model, the training module is configured to obtain a plurality of sets of different sample data and a preset exit probability corresponding to each set of sample data, where the sample data includes the operation data and/or the user source; inputting each group of the sample data into the prediction model to obtain corresponding exit probability; and when the difference value between the exit probability corresponding to the multiple groups of sample data and the preset exit probability is smaller than the preset value, obtaining a trained prediction model.

And training the prediction model to obtain a trained prediction model, and inputting the obtained operation data and/or the user source into the trained prediction model to predict the probability of the user exiting the target application program, thereby obtaining the exit probability.

Further, an exit threshold is pre-stored, the operation data of the user includes operation steps and time sequences between the operation steps, and the pushing module 530 is further configured to determine whether the exit probability is greater than or equal to the exit threshold; if yes, pushing the content related to the last operation step in the time sequence.

By judging the size relationship between the exit probability and the exit threshold, when the exit probability is greater than or equal to the exit threshold, the content related to the last operation step in the time sequence is pushed, so that the content which the user is interested in can be pushed more accurately, the attraction of the recommended content to the user is enhanced, and the viscosity of the user is improved.

Further, a plurality of exit threshold ranges are prestored, and the pushing module 530 is further configured to determine the exit threshold range where the exit probability is located; obtaining push content according to the exit threshold range of the exit probability

Further, the pre-stored correspondence table includes a correspondence between an exit threshold range and a category of the pushed content, and the pushing module 530 is further configured to search the correspondence table, and determine the category of the pushed content corresponding to the exit threshold range where the exit probability is located; pushing the push content associated with the category.

The content is pushed according to the exit threshold range and the corresponding relation between the exit threshold range and the category of the pushed content by determining the exit threshold range where the exit probability is located, so that the content which is interested by the user can be pushed more accurately, the attraction of the recommended content to the user is enhanced, and the viscosity of the user is improved.

Further, the pushing module 530 is further configured to obtain multiple sets of different operation data and/or user sources, where operation steps in the different operation data are different, or time sequences between the operation steps are different; inputting the multiple groups of different operation data and/or user sources into a trained prediction model to obtain multiple corresponding exit probabilities; an exit threshold or a range of exit thresholds is determined from the plurality of exit probabilities.

When the exit threshold or the exit threshold range is determined, a plurality of groups of different operations and user sources can be input into the prediction model to obtain a plurality of exit probabilities, and the exit threshold or the exit threshold range is determined according to the plurality of exit probabilities, so that the accuracy of the exit threshold or the exit threshold range is improved, the content is pushed more timely, and the viscosity of the user is improved.

The content recommendation device 500 provided in this embodiment of the application can implement each process of implementing the content recommendation method by the server in the method embodiments of fig. 2 to fig. 7, and for avoiding repetition, details are not described here again.

An embodiment of the present application provides a server, which includes a processor and a memory, where the memory stores at least one instruction, at least one program, a code set, or a set of instructions, and the at least one instruction, the at least one program, the code set, or the set of instructions are loaded and executed by the processor to implement the content recommendation method provided in the above method embodiment.

The memory may be used to store software programs and modules, and the processor may execute various functional applications and information feedback by operating the software programs and modules stored in the memory. The memory can mainly comprise a program storage area and a data storage area, wherein the program storage area can store an operating system, application programs needed by functions and the like; the storage data area may store data created according to use of the apparatus, and the like. Further, the memory may include high speed random access memory, and may also include non-volatile memory, such as at least one magnetic disk storage device, flash memory device, or other volatile solid state storage device. Accordingly, the memory may also include a memory controller to provide the processor access to the memory.

Fig. 9 is a block diagram of a hardware structure of a server of a content recommendation method according to an embodiment of the present application. As shown in fig. 9, the server 600 may have a relatively large difference due to different configurations or performances, and may include one or more processors (CPUs) 610 (the processors 610 may include but are not limited to Processing devices such as a microprocessor MCU or a programmable logic device FPGA), a memory 630 for storing data, and one or more storage media 620 (e.g., one or more mass storage devices) for storing applications 623 or data 622. Memory 630 and storage medium 620 may be, among other things, transient or persistent storage. The program stored in the storage medium 620 may include one or more modules, each of which may include a series of instruction operations for the electronic device. Further, the processor 610 may be configured to communicate with the storage medium 620 to execute a series of instruction operations in the storage medium 620 on the server 600. The server 600 may also include one or more power supplies 660, one or more wired or wireless network interfaces 650, one or more input-output interfaces 640, and/or one or more operating systems 621, such as Windows Server, MacOSXTM, UnixTM, LinuxTM, FreeBSDTM, and so forth.

The input/output interface 640 may be used to receive or transmit data via a network. Specific examples of the network described above may include a wireless network provided by a communication provider of the server 600. In one example, i/o Interface 640 includes a Network adapter (NIC) that may be coupled to other Network devices via a base station to communicate with the internet. In one example, the input/output interface 640 may be a Radio Frequency (RF) module, which is used for communicating with the internet in a wireless manner.

It will be understood by those skilled in the art that the structure shown in fig. 9 is only an illustration, and is not intended to limit the structure of the server. For example, server 600 may also include more or fewer components than shown in FIG. 9, or have a different configuration than shown in FIG. 9.

The embodiment of the present application further provides a computer-readable storage medium, where a computer program is stored on the computer-readable storage medium, and when the computer program is executed by a processor, the computer program implements the processes of the content recommendation method in the embodiment, and can achieve the same technical effects, and in order to avoid repetition, the details are not repeated here. The computer-readable storage medium may be a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk.

It should be noted that, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element.

Through the above description of the embodiments, those skilled in the art will clearly understand that the method of the above embodiments can be implemented by software plus a necessary general hardware platform, and certainly can also be implemented by hardware, but in many cases, the former is a better implementation manner. Based on such understanding, the technical solutions of the present invention may be embodied in the form of a software product, which is stored in a storage medium (such as ROM/RAM, magnetic disk, optical disk) and includes instructions for enabling a terminal (such as a mobile phone, a computer, a server, an air conditioner, or a network device) to execute the method according to the embodiments of the present invention.

While the present invention has been described with reference to the embodiments shown in the drawings, the present invention is not limited to the embodiments, which are illustrative and not restrictive, and it will be apparent to those skilled in the art that various changes and modifications can be made therein without departing from the spirit and scope of the invention as defined in the appended claims.

Claims

1. A method for recommending content, the method comprising:

acquiring user data in a target application program collected based on the embedded point information, wherein the user data comprises operation data and/or a user source of the target application program used by a user at this time;

obtaining exit probability according to the operation data of the user and/or a user source, wherein the exit probability is the probability that the user corresponding to the user data exits the target application program;

and pushing content according to the exit probability and the operation data.

2. The method of claim 1, wherein obtaining the exit probability according to the user operation data and/or the user source comprises:

and inputting the operation data and/or the user source into a pre-trained prediction model to obtain the exit probability, wherein the prediction model is used for outputting the corresponding exit probability according to the input operation data and/or the user source.

3. The method of claim 2, wherein before inputting the user's operational data and/or user source into the pre-trained predictive model, further comprising:

acquiring a plurality of groups of different sample data and a preset exit probability corresponding to each group of sample data, wherein the sample data comprises operation data and/or a user source;

inputting each group of the sample data into the prediction model to obtain corresponding exit probability;

and when the difference value between the exit probability corresponding to the multiple groups of sample data and the preset exit probability is smaller than the preset value, obtaining a trained prediction model.

4. The method according to any one of claims 1 to 3, wherein an exit threshold is pre-stored, the operation data of the user includes operation steps and time sequences between the operation steps, and the pushing content according to the exit probability and the operation data includes:

judging whether the exit probability is greater than or equal to the exit threshold value;

if yes, pushing the content related to the last operation step in the time sequence.

5. The method according to any one of claims 1-3, wherein a plurality of exit threshold ranges are pre-stored, and wherein pushing content according to the exit probability and the operation data comprises:

determining an exit threshold range in which the exit probability is located;

and acquiring the push content according to the exit threshold range of the exit probability.

6. The method according to claim 5, wherein a correspondence table is stored in advance, the correspondence table includes a correspondence between an exit threshold range and a category of the push content, and the obtaining the push content according to the exit threshold range where the exit probability is located includes:

searching the corresponding relation table, and determining the category of the push content corresponding to the exit threshold range in which the exit probability is located;

pushing the push content associated with the category.

7. The method of claim 4 or 5, wherein before pushing content according to the exit probability and the operation data, further comprising:

acquiring a plurality of groups of different operation data and/or user sources, wherein the operation steps in the different operation data are different, or the time sequences among the operation steps are different;

inputting the multiple groups of different operation data and/or user sources into a trained prediction model to obtain multiple corresponding exit probabilities;

an exit threshold or a range of exit thresholds is determined from the plurality of exit probabilities.

8. A content recommendation apparatus, characterized in that the apparatus comprises:

the system comprises a first acquisition module, a second acquisition module and a third acquisition module, wherein the first acquisition module is used for acquiring user data in a target application program collected based on embedded point information, and the user data comprises operation data and/or a user source of the target application program used by a user at this time;

a second obtaining module, configured to obtain an exit probability according to the operation data of the user and/or a user source, where the exit probability is a probability that a user corresponding to the user data exits the target application program;

and the pushing module is used for pushing the content according to the exit probability and the operation data.

9. A server, characterized in that the server comprises:

one or more processors;

a memory electrically connected with the one or more processors;

one or more applications, wherein the one or more applications are stored in the memory and configured to be executed by the one or more processors, the one or more applications configured to perform the method of any of claims 1-7.

10. A computer-readable storage medium, having stored thereon program code that can be invoked by a processor to perform the method according to any one of claims 1 to 7.