WO2018024222A1 - Application recommendation method, device, and server - Google Patents

Abstract

An application recommendation method, device, and server. The method comprises: acquiring an application list of each of users and an area where each of the users is located (S400); computing, according to the application list of each of the users and the area where each of the users is located, a uniqueness indication corresponding to each of applications in each area (S500); and determining, on the basis of the uniqueness indication corresponding to each of the applications in each area, an application to be recommended with respect to each area. The method is adopted to automatically and effectively select an application most unique in each area without manual intervention, implementing geographical location-based personalized recommendation, more closely meeting needs of users.

Description

Application recommendation method, device and server Technical field

The present invention relates to the field of Internet technologies, and in particular, to an application recommendation method, apparatus, and server.

Background technique

With the rapid development of mobile terminal technology and network technology, more and more users choose to download application software on mobile terminals such as mobile phones or tablet computers, and various application acquisition platforms designed for mobile terminal users, such as application stores, should also be shipped. Born.

Currently, the method of recommending an app by an app store is usually to calculate the download amount of each app in the entire store, sort the app by the download amount, and then recommend the top-ranked app to the user, that is, the app recommendation based on the leaderboard. However, this method will recommend the same application to all users or a certain type of user. The application recommended by each user is the same, not necessarily what the user really wants, ignoring the individual differences of different users' application requirements. The accuracy of the application recommendation is not high, and the user's satisfaction with the recommended application is not high, thereby affecting the software download amount and the user experience.

Summary of the invention

In view of this, an object of the embodiments of the present invention is to provide an application recommendation method, apparatus, and server to solve the above problems.

In order to achieve the above object, the technical solution adopted by the embodiment of the present invention is as follows:

In a first aspect, an embodiment of the present invention provides an application recommendation method, where the method includes: acquiring an application list of each user and an area where each user is located; and according to the application list of each user and the area where each user is located And respectively calculating the uniqueness indicator values corresponding to the respective applications in the respective regions; determining the to-be-recommended applications based on the uniqueness indicator values of the respective applications in the respective regions.

In a second aspect, an embodiment of the present invention provides an application recommendation device, where the device includes: a first acquisition module, configured to acquire an application list of each user and an area where each user is located; and a calculation module, configured to Determining, by the application list of each user, the uniqueness indicator value corresponding to each application in each area, and determining a module, configured to determine, according to the uniqueness indicator value of each application in each area, each area Recommended application.

In a third aspect, an embodiment of the present invention provides a server, where the server includes a memory and a processor, the memory is coupled to the processor, and the memory stores an instruction when the When the instructions are executed by the processor, the processor is configured to: the first obtaining module is configured to obtain an application list of each user and an area where the respective users are located; and a computing module, configured to be used according to the application of each user The list and the area where the user is located respectively calculate the uniqueness indicator values corresponding to the respective applications in the respective areas; and the determining module is configured to determine the to-be-recommended application based on the uniqueness indicator value of each application in the respective areas.

In a fourth aspect, an embodiment of the present invention provides an application recommendation method, where the method includes:

Obtaining a feature vector of the user according to the historical application list;

And predicting a prediction model of each label corresponding to the application according to the feature vectors of the plurality of users; and

A recommendation application sequence of the user is generated based on a feature vector of the user and a prediction model of the respective tags.

In a fifth aspect, an embodiment of the present invention provides an application recommendation device, where the device includes:

a feature vector generating module, configured to calculate a feature vector of the user according to a historical application list of the user; and

a prediction model generating module, configured to train, according to a plurality of feature vectors of the user, a prediction model of each tag corresponding to the application; and

A recommendation application sequence generation module is configured to generate a recommended application sequence of the user based on a feature vector of the user and a prediction model of the respective tags.

In a sixth aspect, an embodiment of the present invention provides a server, including a memory, a processor, and an application recommendation device. The application recommendation device is installed or stored in the memory, and the processor controls execution of each functional module of the application recommendation device.

Compared with the prior art, an application recommendation method, device, and server provided by the embodiments of the present invention calculate the uniqueness index values corresponding to each application in each area, and according to the ranking results of the uniqueness index values, respectively The area selects a preset number of applications as the recommended application corresponding to each area, so that the recommended application corresponding to the area where the user to be recommended is located is returned to the user terminal of the user to be recommended by obtaining the area where the user to be recommended is located. This method can automatically and effectively screen the most unique applications in each region without manual intervention, and realize personalized recommendations based on geographic location, which is more in line with user needs.

Compared with the prior art, the application recommendation method and apparatus provided by the embodiments of the present invention construct a feature vector of the user based on the usage habit of the user; and generate a recommended application sequence of the user by using the feature vector of the user. When the user's recommended application sequence is generated, the user's previous usage habits are considered. When the application recommendation is made, the application that has a high correlation with the user's previous usage habits is recommended to the user, thereby implementing personalized recommendation of the application between different users, and the recommendation is accurate. The high rate helps to increase the number of app downloads and enhance the user experience.

In order to make the above objects, features and advantages of the present invention more comprehensible, The embodiment and the accompanying drawings are described in detail below.

DRAWINGS

In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the embodiments will be briefly described below. It should be understood that the following drawings show only certain embodiments of the present invention, and therefore It should be seen as a limitation on the scope, and those skilled in the art can obtain other related drawings according to these drawings without any creative work.

FIG. 1 is a schematic diagram of a user terminal interacting with a server according to an embodiment of the present invention.

FIG. 2 is a schematic structural diagram of a server according to an embodiment of the present invention.

FIG. 2b is another schematic structural diagram of a server according to an embodiment of the present invention.

FIG. 3 is a flowchart of an application recommendation method according to an embodiment of the present invention.

FIG. 4 is a detailed flowchart of step S500 in an application recommendation method according to an embodiment of the present invention.

FIG. 5 is a detailed flowchart of step S600 of an application recommendation method according to an embodiment of the present invention.

FIG. 6 is a detailed flowchart of step S620 of an application recommendation method according to an embodiment of the present invention.

FIG. 7 is a partial flowchart of an application recommendation method according to an embodiment of the present invention.

FIG. 8 is a structural block diagram of an application recommendation apparatus according to an embodiment of the present invention.

FIG. 9 is a functional block diagram of the application recommendation apparatus shown in FIG. 2b according to an embodiment of the present invention.

FIG. 10 is a block diagram of another functional module of the application recommendation device shown in FIG. 2b according to an embodiment of the present invention.

FIG. 11 is a specific flowchart of an application recommendation method according to an embodiment of the present invention.

FIG. 12 is a schematic flow chart of the sub-steps included in step S1130 in FIG. 11.

FIG. 13 is a specific flowchart of another application recommendation method according to an embodiment of the present invention.

detailed description

The technical solutions in the embodiments of the present invention are clearly and completely described in the following with reference to the accompanying drawings in the embodiments of the present invention. It is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. The components of the embodiments of the invention, which are generally described and illustrated in the figures herein, may be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the invention in the claims All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative efforts are protected by the present invention. The scope.

It should be noted that similar reference numerals and letters indicate similar items in the following figures, and therefore, once an item is defined in a drawing, it is not necessary to further define and explain it in the subsequent drawings. Also, in the description of the present invention, the terms "first", "second", and the like are used merely to distinguish a description, and are not to be construed as indicating or implying a relative importance.

FIG. 1 is a schematic diagram of interaction between a server 200 and a user terminal 100 according to an embodiment of the present invention. The server 200 is in communication with one or more user terminals 100 over a network 300 for data communication or interaction. The server 200 can be a web server, a database server, or the like. The user terminal 100 may be a personal computer (PC), a tablet computer, a smart phone, a personal digital assistant (PDA), an in-vehicle device, a wearable device, or the like. The network 300 can be, but is not limited to, a wired network or a wireless network.

As shown in FIG. 2a, it is a block schematic diagram of the server 200. The server 200 includes a memory 201, a processor 202, and a network module 203.

The memory 201 can be used to store software programs and modules, such as the application recommendation method and the program instructions/modules corresponding to the device in the embodiment of the present invention. The processor 202 performs various function applications and data processing by executing software programs and modules stored in the memory 201, that is, implementing the application recommendation method in the embodiment of the present invention. Memory 201 can include high speed random access memory and can also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid state memory. Further, the software program and the module in the foregoing memory 201 may further include an operating system 221 and a service module 222. The operating system 221, which may be LINUX, UNIX, WINDOWS, may include various software components and/or drivers for managing system tasks (eg, memory management, storage device control, power management, etc.), and may be associated with various hardware or Software components communicate with one another to provide an operating environment for other software components. The service module 222 runs on the basis of the operating system 221, and listens for requests from the network through the network service of the operating system 221, completes corresponding data processing according to the request, and returns the processing result to the client. That is, the service module 222 is configured to provide network services to the client.

The network module 203 is configured to receive and transmit network signals. The above network signal may include a wireless signal or a wired signal.

Please refer to FIG. 2b, which is another block diagram of the server 200 shown in FIG. 1. The server 200 includes an application recommendation device 210, a memory 211, a processor 212, and a communication unit 213.

The components of the memory 211, the processor 212, and the communication unit 213 are electrically connected directly or indirectly to each other to implement data transmission or interaction. For example, the components can be electrically connected to one another via one or more communication buses or signal lines.

The memory 211 may be, but not limited to, a random access memory (Random). Access Memory, RAM, Read Only Memory (ROM), Programmable Read-Only Memory (PROM), Erasable Programmable Read-Only Memory (EPROM), Electric Erasable Programmable Read-Only Memory (EEPROM), etc. The memory 211 is used to store a program, and the processor 212 executes the program after receiving an execution instruction. The communication unit 213 is configured to establish a communication connection between the server 200 and the user terminal 100 through the network 300, and is used to receive and send data through the network 300.

The application recommendation device 210 includes at least one software function module that can be stored in the memory 211 or is solidified in an operating system (OS) of the server 200 in the form of software or firmware. The processor 212 is configured to execute an executable module stored in the memory 211, such as a software function module, a computer program, and the like included in the application recommendation device 210. In this embodiment, the application recommendation device 210 provides the server 200 with a personalized application recommendation service for different users, and the specific method is described in detail later.

It will be understood that the structure shown in Fig. 2a or Fig. 2b is merely illustrative, and the server 200 may further include more or less components than those shown in Fig. 2a or Fig. 2b, or have the same as shown in Fig. 2a or Fig. 2b. Different configurations. The components shown in Figure 2a or Figure 2b can be implemented in hardware, software, or a combination thereof. In addition, the server in the embodiment of the present invention may further include multiple servers with different functions.

In the embodiment of the present invention, the client terminal 100 is installed with a client, and the client may be a third-party application software, which corresponds to the server (Server) and follows the same set of data protocols, so that the server and the client can mutually Analyze the data of the other party and provide the user with the application recommendation service.

FIG. 3 is a flowchart of an application recommendation method according to an embodiment of the present invention. Referring to FIG. 3, this embodiment describes a process flow of a server, where the method includes:

Step S400: Obtain an application list of each user and an area where the respective users are located.

In one example, the user's application list may be implemented in various ways, such as an application download category, an application installation list, an application browsing list, or an application update list, but is not limited to the embodiment. The user's application list can be recorded in the user log. The user's area can be the city where the user is located, or the province where the user is located. The area where the user is located can be obtained through the location information of the user terminal of the user.

For example, an application installation list of the user's most recent time may be collected from the user log, such as an application installation list for the most recent week or the most recent month. Preferably, the system pre-installed application can be excluded from the application installation list, the user-defined installation application is obtained, and the user is The custom installed app serves as a list of user action apps. Then, the user's current city is obtained through the location information of the user terminal of the user, and the user, the city, and the application triplet are obtained, as shown in Table 1.

Table 1

user	city	Application name
A	Guangzhou	Application 1
A	Guangzhou	Application 2
A	Guangzhou	Application 3
A	Guangzhou	Application 4
B	Beijing	Application 1
B	Beijing	Application 5

Step S500: Calculate, according to the application list of each user and the area where each user is located, a uniqueness indicator value corresponding to each application in each area.

In an example, the implementation manners of calculating the uniqueness indicator values corresponding to the respective applications in the respective areas may be various according to the application list of the respective users and the area of the respective users, and one of the embodiments is described below. However, it will be understood that it is not limited to such an embodiment.

Referring to FIG. 4, step S500 may include:

In step S510, the standard deviation of each application in all areas is calculated according to the number of users of each application in each area and the average number of users of each application in all areas.

In an example, the number of users of each application in each area may be calculated according to the application list of each user and the area where each user is located.

For example, suppose that the application list of each user is obtained and the area of each user is as shown in Table 2 below:

Table 2

user	city	Application name
A	Guangzhou	Application 1
A	Guangzhou	Application 2
B	Guangzhou	Application 1
B	Guangzhou	Application 3
C	Beijing	Application 1
C	Beijing	Application 3

The number of users of each application in each area is represented by u _i,j , where i represents a city and j represents an application. Then the value of i is (Guangzhou, Beijing), and the value of j is (Application 1, Application 2, Application 3). The number of users for each application in each region is as follows:

u (Guangzhou, application 1) = 2;

u (Guangzhou, application 2) = 1;

u (Guangzhou, application 3)=1;

u (Beijing, application 1) = 1;

u (Beijing, application 2) = 0;

u (Beijing, application 3)=1.

Next, the average number of users for each application in all regions is calculated separately.

If the number of users of each application in each area is u _i,j , where i is the city and j is the application, the average number of users in each area is:

Where N is the number of regions.

In the example above, N=2, respectively, Guangzhou and Beijing, the value of i is (Guangzhou, Beijing), and the value of j is (Application 1, Application 2, Application 3), then each application is in all areas. The average number of users is as follows:

Application 1 = [u (Guangzhou, Application 1) + u (Beijing, Application 1)] / 2 = (2 + 1) / 2 = 1.5;

Application 2 = [u (Guangzhou, Application 2) + u (Beijing, Application 2)] / 2 = (1 + 0) / 2 = 0.5;

Application 2 = [u (Guangzhou, Application 3) + u (Beijing, Application 3)] / 2 = (1 + 1) / 2 = 1.

When calculating the number of users of each application in each area and the average number of users in each area in each area, the standard deviation of each application in all areas can be calculated according to the following formula:

Where u _i,j is the number of users of each application in each area, i is the city, j is the application, μ _j is the average number of users in each area, and N is the number of areas.

In step S520, the uniqueness index values corresponding to the respective applications in the respective regions are respectively calculated according to the number of users of each application in the respective regions and the standard deviation of the respective applications in all regions.

Preferably, the uniqueness indicator value corresponding to each application in each area is a ratio of the number of users of each application in the respective areas to the standard deviation of the respective applications in all areas. That is, the uniqueness index value can be calculated according to the following formula:

_Wherein, u _{i, j} is the number of users in each area of each application, i denotes the city, j represents the application, σ _j for each application a standard deviation of all regions.

It can be understood that the distribution of the number of users in an application is approximately Gaussian. The following formula:

In the Gaussian distribution, the standard deviation σ of the distribution reflects the degree of concentration of the distribution, that is, the more concentrated the distribution, the smaller the σ. Therefore, the uniqueness indicator

For the same city, the larger the value of S, the more unique the application. That is, the app is the most commonly used app for users in the city and is a unique application for the city.

Referring to FIG. 3, after step S500 is performed, step S600 is continued.

Step S600: Determine an application to be recommended for each area based on a uniqueness indicator value of each application in each of the areas.

As an implementation manner, referring to FIG. 5, the step S600 may include:

Step S610: Sort each application in each area according to the uniqueness indicator value to obtain a candidate application set to be recommended for each area.

For the city i, the uniqueness indicators S _{i,j of} all the applications j are sorted in descending order, and the top K applications ranked in the respective regions are selected as the recommended applications corresponding to each region.

For example, suppose that each application in each region is sorted according to the uniqueness indicator value, as shown in Table 3 below:

table 3

city	Application name	Uniqueness index
Guangzhou	Application 1	1.00294399261
Guangzhou	Application 2	1.00283753872
Guangzhou	Application 3	1.00281596184
Guangzhou	Application 4	1.00281572342
Guangzhou	Application 5	1.00280892849
Beijing	Application 2	1.00267326832
Beijing	Application 3	1.00266420841
Beijing	Application 4	1.00261163712
Beijing	Application 5	1.00259315968
Beijing	Application 1	1.00248968601

Assuming that the first preset number is 3, three applications are selected for each region as the to-be-recommended application set corresponding to each region. The set of applications to be recommended for each region is shown in Table 4 below:

Table 4

city	Application name	Uniqueness index
Guangzhou	Application 1	1.00294399261
Guangzhou	Application 2	1.00283753872
Guangzhou	Application 3	1.00281596184
Beijing	Application 2	1.00267326832
Beijing	Application 3	1.00266420841
Beijing	Application 4	1.00261163712

Step S620: Determine a to-be-recommended application of each area from the to-be-recommended application candidate set of each area.

Preferably, referring to FIG. 6, the step S620 may include:

Step S621: Calculate the number of users corresponding to each application in the to-be-recommended application set corresponding to each area.

The number of install/download users per application may be calculated based on log data of the installed/downloaded application of the user; the application of the set of applications to be recommended for each region is associated with the number of install/download users of each application , the data after the association, as shown in Table 5 below:

table 5

city	Application name	Uniqueness index	User number
Guangzhou	Application 1	1.00294399261	10000
Guangzhou	Application 2	1.00283753872	8888
Guangzhou	Application 3	1.00281596184	15247
Beijing	Application 2	1.00267326832	9856
Beijing	Application 3	1.00266420841	12003
Beijing	Application 4	1.00261163712	10036

In step S622, the applications in the application group to be recommended corresponding to each area are sorted according to the number of users, and the application to be recommended for each area is obtained.

The application in the to-be-recommended application set corresponding to each area may be sorted in descending order according to the number of users, and the first second preset number application ranked in the previous area is selected as the recommended application for each area. The second preset number may be less than or equal to the first preset number.

In the example above, after reordering, the list of recommended applications for each area is shown in Table 6 below:

Table 6

city	Application name	Recommended sequence
Guangzhou	Application 3	1
Guangzhou	Application 1	2
Guangzhou	Application 2	3
Beijing	Application 4	1
Beijing	Application 3	2
Beijing	Application 2	3

If the second preset number is 2, the applications recommended to users in Guangzhou are Application 3 and Application 2, and the applications recommended to users in Beijing are Application 4 and Application 3.

It can be understood that after selecting the corresponding recommended application for each area, different applications can be recommended to different regional users respectively.

Referring to FIG. 7, after step S600, the method may further include the following steps:

Step S700, obtaining an area where the user to be recommended is located;

Step S800: Query the recommended application corresponding to the area where the user to be recommended is located;

Step S900: Return, to the user terminal of the user to be recommended, a recommended application corresponding to the area where the user to be recommended is located.

An application recommendation method provided by an embodiment of the present invention calculates a uniqueness indicator value corresponding to each application in each area, and selects a preset number of applications for each area according to the ranking result of the uniqueness indicator value. The recommended application corresponding to the area, so that the recommended application corresponding to the area where the user to be recommended is located is returned to the user terminal of the user to be recommended. This method can automatically and effectively screen the most unique applications in each region without manual intervention, and realize personalized recommendation based on geographic location, thereby enhancing the big data perception of the product. In addition, because the recommended application combines the geographic location of the user, the user download amount can be increased to some extent. Further, after sorting according to the uniqueness index value, the application according to the application group to be recommended is sorted according to the number of users, and the second preset number of applications are selected for each region as the recommended application corresponding to each region. In this way, the recommendation application of each region is filtered by the uniqueness index and the number of users of the application, so that the recommended application corresponding to each region finally selected is more accurate.

FIG. 8 is a schematic diagram of functional modules of an application recommendation device 210 according to an embodiment of the present invention. The application recommendation device 210 includes a first acquisition module 211, a calculation module 212, and a determination module 213.

The first obtaining module 211 is configured to obtain an application list of each user and an area where the respective users are located.

The calculation module 212 is configured to calculate, according to the application list of each user and the area where each user is located, a uniqueness indicator value corresponding to each application in each area.

Preferably, the calculation module 212 includes a first calculation sub-module 2121 and a second calculation sub-module 2122. The first calculation sub-module 2121 is configured to calculate a standard deviation of each application in all areas according to the number of users of each application in each area and the average number of users in each area in each area; the second calculation The sub-module 2122 is configured to separately calculate the uniqueness indicator values corresponding to the respective applications in the respective regions according to the number of users of each application in the respective regions and the standard deviation of the respective applications in all regions.

In one example, the uniqueness indicator values corresponding to the respective applications in the respective regions are the ratio of the number of users of each application in the respective regions to the standard deviation of the respective applications in all regions.

The determining module 213 is configured to determine, according to the uniqueness indicator value of each application in the respective regions, an application to be recommended for each region.

Preferably, the determining module 213 includes a sorting sub-module 2131 and a selected sub-module 2132. The sorting sub-module 2131 is configured to sort each application in each area according to the uniqueness index value to obtain a candidate application candidate set for each area; the selected sub-module 2132 is configured to use each The candidate application candidates to be recommended in the region determine the application to be recommended for each region.

Preferably, the selected sub-module 2132 is configured to calculate a number of users corresponding to each application in the to-be-recommended application set corresponding to each area, and apply the application to the recommended application set corresponding to each area according to the user. Sort the numbers to get the recommended application for each area.

Preferably, the device further includes a second obtaining module 214, a query module 215 and a return module 216. The second obtaining module 214 is configured to obtain an area where the user to be recommended is located; the querying module 215 is configured to query a recommended application corresponding to the area where the user to be recommended is located; and the returning module 216 is configured to The user terminal that describes the recommended user returns the recommended application corresponding to the area where the user to be recommended is located.

Each of the above modules may be implemented by software code. In this case, each of the above modules may be stored in the memory 201 of the server 200 shown in FIG. 2a. The above modules can also be implemented by hardware such as an integrated circuit chip.

Please refer to FIG. 9, which is a functional block diagram of the application recommendation device 210 shown in FIG. 2b according to a preferred embodiment of the present invention. The application recommendation device 210 includes a feature vector generation module 901, a prediction model generation module 902, and a recommendation application sequence generation module 903.

The feature vector generation module 901 is configured to calculate a feature vector of the user according to a history application list of the user.

In this embodiment, the history application list may include an installed application list of the user terminal 100, a download application list in a preset time (for example, a recent time period), and a preset. A list of browsing applications and a list of updated applications within a preset time.

The process of the feature vector generation module 901 calculating the feature vector of the user according to the history application list of the user is specifically described below.

First, the feature vector generation module 901 calculates a score of each application in the history application list.

The feature vector generation module 901 acquires a list of installed applications for each user, a list of downloaded applications within a preset time, a list of browsing applications within a preset time, and an updated application list within a preset time. Calculate the scores for all the apps in each app list. The score calculation method is as follows:

For the updated app list, the applied score is:

Update_app_score _i =update_app _i ×w _update ;

For browsing the list of apps, the app’s score is:

View_app_score _i =view_app _i ×w _viewe ;

For the list of downloaded apps, the applied score is:

Download_app_score _i =download_app _i ×w _download ;

For the list of installed apps, the app’s score is:

Install_app_score _i =install_app _i ×w _instadll .

Among them, update_app _i , view_app _i , download_app _i and install_app _i respectively represent the number of updates, browsing times, download times and installation times of the application i; w _update , w _view , w _download and w _install respectively indicate the weight of the update times and the number of views The weight, the weight of the downloads, and the weight of the number of installs can be preset.

Taking user i as an example, the score distribution of the application of user i refers to Table 7.

Table 7

The feature vector generation module 901 assigns a label to each application in the history application list.

Specifically, the feature vector generation module 901 queries the label corresponding to the application from the mapping table of the application and the label, and expands the table 7, and labels each application to obtain the table 8.

Table 8

The feature vector generation module 901 calculates a probability distribution value of a tag of a single user.

Specifically, the feature vector generation module 901 sums the application scores under the same label of each user, and obtains various types of label scores for each user. Please refer to Table 9.

Table 9

Each type of tag score of each user is normalized to obtain a tag probability distribution of the user, and the probability distribution value of the tag j of the user i is represented as p(c _j |u _i ), refer to Table 10.

Table 10

The feature vector generation module 901 calculates a probability distribution value of tags of all users.

Specifically, the feature vector generation module 901 sums and counts each tag score for all users, and normalizes to a tag probability distribution value of all users, and the probability distribution value of each tag j of all users is represented as p(c _j ), please refer to Table 11 for details.

Table 11

label	p(c j )
play music	3.03%
Online shopping	16.16%
Social	15.29%
.......	0.15%
.......	0.42%
.......	0.09%
.......	0.30%
.......	0.54%

The feature vector generation module 901 obtains a feature vector of the user according to a probability distribution value of the label of the single user and a probability distribution value of the labels of all users.

Specifically, the feature vector generation module 901 calculates a feature vector of each user, and the formula for calculating the feature vector of the user i is: V _i =[v _i,1 , v _i,2 , . . . , v _i,j ,...,v _i,N ], where:

N represents the number of labels. Refer to Table 12 for the relationship between the feature values v _i,j and the tags in the feature vector of the user i.

Table 12

label	v i,j
Social	2.18
Online shopping	2.06
play music	11.00

In this embodiment, machine learning can also be adopted based on the feature vector V _i =[v _i,1 , v _i,2 , . . . , v _i,j , . . . , v _i,N ] of the user _i. The method re-obtains a new feature vector to characterize the characteristics of user i.

The prediction model generating module 902 is configured to train, according to the feature vectors of the plurality of users, a prediction model of each tag corresponding to the application.

In this embodiment, the prediction model generation module 902 first generates a training sample set applied under each label, and the training sample set generation method is as follows.

If the feature vector of user i generated in t days is V _i =[v _i,1 , v _i,2 ,...,v _i,j ,...,v _i,N ], and at t+ After downloading the application labeled L for 1 day, the training sample data of the user is S _i =[v _i,1 ,v _i,2 ,...,v _i,j ,...,v _i,N , 1]; otherwise, it is denoted as S _i =[v _i,1 ,v _i,2 ,...,v _i,j ,...,v _i,N ,0]. Among them, the last dimension is the training goal. A training sample is generated for all users to obtain a training sample set {S _i } including all user training samples.

The prediction model generation module 902 trains the training sample set {S _i } of all users under the label L by using a decision tree model to obtain a decision tree prediction model M _L for the label L.

The prediction model generation module 902 then trains all the tags using the above method to obtain a prediction model set {M _L } of the prediction models including all tags.

The recommended application sequence generating module 903 is configured to generate a recommended application sequence of the user based on a feature vector of the user and a prediction model of the respective tags.

Referring to FIG. 10, in the embodiment, the recommended application sequence generation module 903 may include a tag sequence generation submodule 9031 and a recommendation application sequence generation submodule 9032.

The specific manner of generating the recommended application sequence of the user may be implemented by using the foregoing sub-module, and the specific implementation manner is as follows.

The tag sequence generation sub-module 9031 is configured to input the feature vectors of the user into the prediction models of the respective tags to obtain the predicted tag sequence of the user.

The tag sequence generation sub-module 9031 inputs the feature vector of each user into the prediction model of each tag to obtain a prediction tag sequence U _i ={L _j } for each user.

The tag sequence generation sub-module 9031 generates a tag sequence set including a prediction tag sequence of all users according to a prediction tag sequence of each user.

The recommended application sequence generation sub-module 9032 is configured to generate a recommended application sequence of the user according to the predicted tag sequence and an application under the predictive tag.

The recommended application sequence generation sub-module 9032 sorts the user's predicted tag sequence. In this embodiment, each tag in the predicted tag sequence is sorted according to the user's Gini value (gini value) of the tag. Specifically, the order can be sorted from small to large according to the Gini value.

The recommended application sequence generation sub-module 9032 sorts all applications under the label to obtain a recommendation sequence applied under the label. In this embodiment, all the applications under the label are sorted according to the number of downloads in the preset time, and specifically, the number of downloads applied in the recommended sequence may be sorted from large to small.

The recommended application sequence generation sub-module 9032 obtains a recommended application sequence for each user according to the sequence number of each tag in each user's predicted tag sequence and the recommended sequence number applied under each tag.

In this embodiment, the recommended application sequence generation submodule 9032 sorts each tag in the predicted tag sequence; and sorts all applications under each tag. A recommended application sequence for each user is calculated according to the ordering of each tag in each user's predicted tag sequence and the ordering of applications under each tag.

Specifically, the recommended application sequence generation sub-module 9032 calculates a recommended application sequence of each user by: serializing each serial number of each label in each user's prediction label sequence. The recommended serial number of the application under the label is multiplied, and the application is sorted according to the product (such as sorting in ascending order) to obtain the recommended application sequence of the user.

The recommended application sequence generation sub-module 9032 further generates a recommended application sequence set including a recommended application sequence of all users according to the recommended application sequence of each user.

Referring to FIG. 10 again, in the embodiment, the application recommendation device 210 may further include an application sequence recommendation module 904, where the application sequence recommendation module 904 is configured to use the recommended application of the user when the user accesses the application store. The application in the sequence is recommended to the user.

Specifically, the application sequence recommendation module 904 may recommend an application ranked first (eg, the top three) in the recommended application sequence to the user for the user to select to download.

Please refer to FIG. 11, which is a flowchart of an information copying method applied to the server 200 shown in FIG. 2b according to a preferred embodiment of the present invention. The method steps of the process-related process definitions may be implemented by the processor 212. The specific flow shown in FIG. 11 will be described in detail below.

Step S1110: Calculate a feature vector of the user according to a history application list of the user.

In this embodiment, the historical application list includes an installed application list, a downloaded application list within a preset time, a browsing application list within a preset time, and an updated application list within a preset time.

The step S1110 includes: calculating a score of each application in the history application list;

Assigning a label to each application in the history application list; calculating a probability distribution value of the label of the individual user; calculating a probability distribution value of the label of all users; and a probability distribution value according to the label of the single user and a probability of labeling of all users The distribution value gets the feature vector of the user.

The process described in the above step S1110 can be performed and implemented by the feature vector generation module 901.

Step S1120: The prediction model of each tag corresponding to the application is obtained according to the feature vectors of the plurality of users.

Specifically, step S1120 includes: generating a training sample set applied under each label. The training sample set is trained using a decision tree model to obtain a prediction model for each tag, and a prediction model set including prediction models of all tags is generated according to the prediction model of each tag.

The process described in the above step S1120 can be performed and implemented by the predictive model generation module 902.

Step S1130: Generate a recommended application sequence of the user based on the feature vector of the user and the prediction model of each tag.

Referring to FIG. 12, in the embodiment, the step S1130 may include sub-step S1131 and sub-step S1132.

Sub-step S1131, the feature vectors of the users are respectively input into a prediction model of each tag to obtain a predicted tag sequence of the user.

The process described in the above sub-step S1131 can be performed and implemented by the tag sequence generation sub-module 9031.

Sub-step S1132, generating a recommended application sequence of the user according to the prediction tag sequence and an application under the prediction tag.

Specifically, the sub-step S1132 includes: sorting each label in the predicted label sequence; sorting all applications under each label; and sorting each label in each of the user's predicted label sequences and under each label The ordering of the applications is calculated to get the recommended application sequence for each user. The label in the predicted label sequence is sorted according to the size of the user's Gini value of the label.

The step of calculating the recommended application sequence of each user may include: obtaining a recommended application sequence of each user according to the serial number of each label in each predicted label sequence of each user and the recommended serial number applied under each label.

The process described in the above sub-step S1132 can be performed and implemented by the recommended application sequence generation sub-module 9032.

Referring to FIG. 13, in the embodiment, the application recommendation method may further include step S1140.

In step S1140, when the user accesses the application store, the application in the recommended application sequence of the user is recommended to the user.

Specifically, the server 200 may recommend an application ranked in the top of the recommended application sequence (for example, the top three) to the user for the user to select to download.

The process described in the above step S1140 can be performed and implemented by the application sequence recommendation module 904.

In summary, the application recommendation method and apparatus provided by the embodiments of the present invention construct a feature vector of the user based on the user's usage habits, and train the label corresponding to the application by using the feature vector of the user to obtain a prediction model set of the label. And then obtaining a prediction tag sequence set according to the feature vector of the user and the prediction model set of the tag; and finally generating a recommended application sequence set of all users according to the prediction tag sequence set. When the user's recommended application sequence is generated, the user's previous usage habits are taken into consideration. When the application recommendation is made, the application that has a high correlation with the user's previous usage habits is recommended to the user, thereby implementing personalized recommendation of the application between different users, and recommending High accuracy, help increase app downloads and improve user experience.

The embodiment of the present invention further provides an application recommendation method, which is applied to a server, and the method includes: calculating a feature vector of the user according to a historical application list of the user; and obtaining an application corresponding according to the feature vector of the plurality of users a prediction model for each of the tags; and a user-based feature vector and a prediction model for the respective tags to generate a recommended application sequence for the user.

Optionally or alternatively, the step of generating the recommended application sequence of the user based on the user-based feature vector and the prediction model of each tag comprises: inputting the feature vector of the user into a prediction model of each tag And obtaining a predicted tag sequence of the user; and generating a recommended application sequence of the user according to the predicted tag sequence and an application under the predictive tag.

Optionally or alternatively, the step of generating the recommended application sequence of the user according to the prediction label sequence and the application under the prediction label comprises: sorting each label in the prediction label sequence; All applications under each label are sorted; and the recommended application sequence of each user is calculated according to the ordering of each label in each user's predicted label sequence and the ordering of applications under each label.

Optionally or alternatively, each of the labels in the predicted tag sequence is sorted according to the size of the user's Gini value of the tag.

Optionally or alternatively, the step of calculating a recommended application sequence of each user according to the ordering of each label in each user's predicted label sequence and the ordering of applications under each label comprises: according to each user The sequence number of each tag in the tag sequence and the recommended sequence number applied under each tag are predicted to obtain a recommended application sequence for each user.

Optionally or alternatively, the method further comprises: recommending an application in the user's recommended application sequence to the user when the user accesses the application store.

Optionally or alternatively, the step of calculating the feature vector of the user according to the history application list of the user comprises: calculating a score of each application in the history application list; and assigning each application in the history application list a label; a probability distribution value of a label of a single user; a probability distribution value of a label of all users; and a feature vector of the user according to a probability distribution value of the label of the single user and a probability distribution value of the labels of all users.

Optionally or alternatively, the historical application list includes an installed application list, a downloaded application list within a preset time, a browsing application list within a preset time, and an updated application list within a preset time.

Optionally or alternatively, the step of training the prediction models of the respective tags corresponding to the application according to the feature vectors of the plurality of users comprises: generating a training sample set applied under each tag; and the training sample set The decision tree model is used to train to obtain a prediction model for each tag, and a prediction model set including prediction models of all tags is generated according to the prediction model of each tag.

The embodiment of the present invention further provides an application recommendation device, which is applied to a server, and the device includes: a feature vector generation module, configured to calculate a feature vector of the user according to a history application list of the user; and a prediction model generation module, configured to: And predicting a prediction model of each label corresponding to the application according to the feature vector of the plurality of users; and a recommendation application sequence generation module, configured to generate the recommended application of the user based on the feature vector of the user and the prediction model of each label sequence.

Optionally or alternatively, the recommended application sequence generating module includes: a tag sequence generating submodule, configured to input the feature vectors of the user into the prediction models of the respective tags to obtain the predicted tag sequence of the user. And a recommendation application sequence generation submodule, configured to generate the recommended application sequence of the user according to the predicted tag sequence and an application under the prediction tag.

Optionally or alternatively, the recommended application sequence generation sub-module obtains a recommended application sequence of each user by: sorting each label in the prediction label sequence; and sorting all applications under each label; And according to the ordering of each label in each user's predicted label sequence and the ordering of the applications under each label, the recommended application sequence of each user is calculated.

Optionally or alternatively, the recommended application sequence generating sub-module sorts each tag in the predicted tag sequence according to the user's Gini value size of the tag.

Optionally or alternatively, the recommended application sequence generation sub-module obtains a recommended application sequence of each user according to a sequence number of each tag in each user's predicted tag sequence and a recommended sequence number applied under each tag.

Optionally or alternatively, the device further includes: an application sequence recommendation module, configured to recommend an application in the recommended application sequence of the user to the user when the user accesses the application store.

Optionally or alternatively, the feature vector generation module obtains a feature vector of the user by: calculating a score of each application in the history application list; assigning a label to each application in the history application list; a probability distribution value of a label of a single user; calculating a probability distribution value of labels of all users; and obtaining a feature vector of the user according to a probability distribution value of the label of the single user and a probability distribution value of labels of all users.

Optionally or alternatively, the historical application list includes an installed application list, a downloaded application list within a preset time, a browsing application list within a preset time, and an updated application list within a preset time.

Optionally or alternatively, the prediction model generation module obtains a prediction model by: generating a training sample set applied under each label; and using the decision tree model for training the training sample set to obtain a prediction of each label A model, and a prediction model set including a prediction model of all tags is generated based on the prediction model of each tag.

An embodiment of the present invention further provides a server, including: a memory; a processor; and an application recommendation device, the device being installed in the memory and including one or more software function modules executed by the processor, The device includes: a feature vector generating module, and calculating a feature vector of the user according to a history application list of the user; and a prediction model generating module, configured to: according to the feature vectors of the plurality of users, obtain a prediction model of each tag corresponding to the application And a recommendation application sequence generation module for generating a recommended application sequence of the user based on the user's feature vector and the prediction model of the respective tags.

In the several embodiments provided in the present application, it should be understood that the disclosed apparatus and method may also be implemented in other manners. The device embodiments described above are merely illustrative, for example, the flowcharts and block diagrams in the figures illustrate the architecture, functionality, and functionality of possible implementations of devices, methods, and computer program products according to various embodiments of the invention. operating. In this regard, each block of the flowchart or block diagram can represent a module, a program segment, or a portion of code that includes one or more of the Executable instructions. It should also be noted that, in some alternative implementations, the functions noted in the blocks may also occur in a different order than those illustrated in the drawings. For example, two consecutive blocks may be executed substantially in parallel, and they may sometimes be executed in the reverse order, depending upon the functionality involved. It is also noted that each block of the block diagrams and/or flowcharts, and combinations of blocks in the block diagrams and/or flowcharts, can be implemented in a dedicated hardware-based system that performs the specified function or function. Or it can be implemented by a combination of dedicated hardware and computer instructions.

In addition, each functional module in each embodiment of the present invention may be integrated to form a separate part, or each module may exist separately, or two or more modules may be integrated to form a separate part.

The functions, if implemented in the form of software functional modules and sold or used as separate products, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present invention, which is essential or contributes to the prior art, or a part of the technical solution, may be embodied in the form of a software product, which is stored in a storage medium, including The instructions are used to cause a computer device (which may be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods described in various embodiments of the present invention. The foregoing storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, and the like. .

It should be noted that, in this context, relational terms such as first and second are used merely to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply such entities or operations. There is any such actual relationship or order between them. Furthermore, the term "comprises" or "comprises" or "comprises" or any other variations thereof is intended to encompass a non-exclusive inclusion, such that a process, method, article, or device that comprises a plurality of elements includes not only those elements but also Other elements, or elements that are inherent to such a process, method, item, or device. An element that is defined by the phrase "comprising a ..." does not exclude the presence of additional equivalent elements in the process, method, item, or device that comprises the element.

The above description is only the preferred embodiment of the present invention, and is not intended to limit the present invention, and various modifications and changes can be made to the present invention. Where in the essence of the invention Any modifications, equivalent substitutions, improvements, etc. made within the principles of God and the principles are intended to be included within the scope of the present invention. It should be noted that similar reference numerals and letters indicate similar items in the following figures, and therefore, once an item is defined in a drawing, it is not necessary to further define and explain it in the subsequent drawings.

The above is only a specific embodiment of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art can easily think of changes or substitutions within the technical scope of the present invention. It should be covered by the scope of the present invention. Therefore, the scope of the invention should be determined by the scope of the claims.

Claims (13)

1. An application recommendation method, wherein the method comprises:

   Obtaining a list of applications of each user and an area where the respective users are located;

   Determining, according to the application list of each user and the area where each user is located, a uniqueness indicator value corresponding to each application in each area;

   The application to be recommended is determined for each area based on the uniqueness indicator value of each application in the respective areas.
2. The method according to claim 1, wherein the calculating the uniqueness indicator value corresponding to each application in each area according to the application list of each user and the area of each user, respectively:

   Calculate the standard deviation of each application in all areas according to the number of users of each application in each area and the average number of users in each area in each area;

   The uniqueness indicator values corresponding to the respective applications in the respective regions are respectively calculated according to the number of users of each application in the respective regions and the standard deviation of the respective applications in all regions.
3. The method according to claim 2, wherein the calculating the standard deviation of each application in all areas according to the number of users of each application in each area and the average number of users of each application in all areas, including:

   The uniqueness indicator value corresponding to each application in each area is a ratio of the number of users of each application in the respective areas to the standard deviation of the respective applications in all areas.
4. The method according to any one of claims 1-3, wherein the determining the to-be-recommended application based on the uniqueness indicator value of each application in the respective regions comprises:

   Sorting each application in each area according to the uniqueness indicator value to obtain a candidate application set to be recommended for each area;

   The to-be-recommended application of each area is determined from the to-be-recommended application candidate set of each of the areas.
5. The method according to claim 4, wherein determining the to-be-recommended application of each area from the to-be-recommended application candidate set comprises:

   Calculating a number of users corresponding to each application in the to-be-recommended application set corresponding to each area;

   The applications in the to-be-recommended application set corresponding to each area are sorted according to the number of users, and the to-be-recommended application of each area is obtained.
6. The method of any of claims 1-5, wherein the method further comprises:

   Obtain the area where the user to be recommended is located;

   Querying the recommended application corresponding to the area where the user to be recommended is located;

   Returning, to the user terminal of the user to be recommended, a recommended application corresponding to the area where the user to be recommended is located.
7. An application recommendation device, wherein the device comprises:

   a first obtaining module, configured to acquire an application list of each user and an area where each user is located;

   a calculation module, configured to calculate, according to the application list of each user and the area where each user is located, a uniqueness indicator value corresponding to each application in each area;

   a determining module, configured to determine an application to be recommended for each region based on a uniqueness indicator value of each application in the respective regions.
8. The apparatus of claim 7, wherein the computing module comprises a first computing sub-module and a second computing sub-module,

   The first calculation sub-module is configured to calculate a standard deviation of each application in all areas according to the number of users of each application in each area and the average number of users of each application in all areas;

   The second calculation sub-module is configured to calculate, according to the number of users of each application in the respective regions and the standard deviation of the respective applications in all regions, the uniqueness index values corresponding to the respective applications in the respective regions.
9. The apparatus according to claim 8, wherein the uniqueness index corresponding to each application in each of the regions is a ratio of a number of users of each application in the respective regions to a standard deviation of the respective applications in all regions.
10. The apparatus according to any one of claims 7-9, wherein the determining module comprises a sorting submodule, and the selected submodule,

    The sorting sub-module is configured to sort each application in each area according to the uniqueness index value, to obtain a candidate application set to be recommended for each area;

    The selected sub-module is configured to determine, from the candidate to be recommended application set of each area, a to-be-recommended application of each area.
11. The device according to claim 10, wherein the selected sub-module is specifically configured to calculate a number of users corresponding to each application in the to-be-recommended application set corresponding to each area; The application in the recommended application set is sorted according to the number of users, and the application to be recommended for each area is obtained.
12. The device according to any one of claims 7-11, wherein the device further comprises a second acquisition module, a query module and a return module,

    The second obtaining module is configured to obtain an area where the user to be recommended is located;

    The querying module is configured to query a recommended application corresponding to the area where the user to be recommended is located;

    The returning module is configured to return a recommended application corresponding to the area where the user to be recommended is located to the user terminal of the user to be recommended.
13. A server, wherein the server includes a memory and a processor, the memory Coupled to the processor, the memory stores instructions that, when executed by the processor, cause the processor to perform the following operations:

    Obtaining a list of applications of each user and an area where the respective users are located;

    Determining, according to the application list of each user and the area where each user is located, a uniqueness indicator value corresponding to each application in each area;

    The application to be recommended is determined for each area based on the uniqueness indicator value of each application in the respective areas.

Images