CN110032678B

CN110032678B - Service resource pushing method and device, storage medium and electronic device

Info

Publication number: CN110032678B
Application number: CN201910276813.6A
Authority: CN
Inventors: 高礼; 李华剑
Original assignee: Tencent Technology Shenzhen Co Ltd
Current assignee: Tencent Technology Shenzhen Co Ltd
Priority date: 2019-04-08
Filing date: 2019-04-08
Publication date: 2023-08-25
Anticipated expiration: 2039-04-08
Also published as: CN110032678A

Abstract

The invention discloses a service resource pushing method and device, a storage medium and an electronic device. Wherein the method comprises the following steps: determining a first group of target accounts related to a first account of a first application in an account set of the first application, wherein a first business resource operated by the first account in the first application is similar to a second business resource operated by the first group of target accounts in the first application; acquiring a second application set of the first group of target accounts operated by the first application; pushing part or all of the second application set to the first account. The invention solves the technical problem of poor pushing efficiency of the service resource pushing method in the related technology.

Description

Service resource pushing method and device, storage medium and electronic device

Technical Field

The present invention relates to the field of computers, and in particular, to a service resource pushing method and device, a storage medium, and an electronic device.

Background

Currently, clients of some applications (e.g., clients of online video media platforms, clients of news reading, clients of web reading, etc.) may provide pushing of other services (e.g., games) while providing mainly one service function (e.g., video media related services, news reading related services, web browsing related services, etc.).

In the related art, when pushing service resources, a pushing manner based on a friend relationship is generally adopted, for example, for the user a, other service resources operated by the friend user B or the user C can be pushed to the user a. However, pushing a service resource to a user based on a friend relationship may not be required by the user, resulting in that the user does not operate the pushed service resource, and thus the service resource pushing efficiency is poor.

Disclosure of Invention

The embodiment of the invention provides a service resource pushing method and device, a storage medium and an electronic device, which at least solve the technical problem that the pushing efficiency of the service resource pushing method in the related technology is poor.

According to an aspect of the embodiment of the present invention, there is provided a service resource pushing method, including: determining a first group of target accounts related to a first account in a first application in an account set of the first application, wherein a first business resource operated by the first account in the first application is similar to a second business resource operated by the first group of target accounts in the first application; acquiring a second application set of the first group of target accounts operated by the first application; pushing part or all of the second application set to the first account.

According to another aspect of the embodiment of the present invention, there is provided a service resource pushing method, including: determining that a display page of a client of a first application logged in by using a first account is slid to a target display area, wherein the target display area is used for displaying one or more second applications pushed to the first account, the one or more applications are part or all of a second application set, the second application set is a set of second applications, a first group of target accounts is operated by the first application, the first group of target accounts is a group of accounts related to the first account, which is determined from an account set of the first application, and first business resources operated by the first account in the first application are similar to second business resources operated by the first group of target accounts in the first application; and pushing the one or more second applications to the first account in the target display area.

According to still another aspect of the embodiment of the present invention, there is further provided a service resource pushing device, including: a first determining unit, configured to determine, in an account set of a first application, a first group of target accounts related to a first account in the account set, where a first service resource operated by the first account in the first application is similar to a second service resource operated by the first group of target accounts in the first application; the acquisition unit is used for acquiring a second application set of the first group of target accounts operated by the first application; and the pushing unit is used for pushing part or all of the second application set to the first account.

According to still another aspect of the embodiment of the present invention, there is further provided a service resource pushing device, including: a second determining unit, configured to determine that a display page of a client of a first application logged in using a first account is slid to a target display area, where the target display area is configured to display one or more second applications pushed to the first account, where the one or more applications are part or all of a second application set, where the second application set is a set of second applications that a first group of target accounts operates through the first application, and the first group of target accounts is a group of accounts related to the first account, which is determined from an account set of the first application, and a first service resource operated by the first account in the first application is similar to a second service resource operated by the first group of target accounts in the first application; and the display unit is used for pushing the one or more second applications to the first account in the target display area.

According to a further aspect of embodiments of the present invention, there is also provided a storage medium storing a computer program arranged to perform the method described above when run.

According to another aspect of the embodiments of the present invention, there is also provided an electronic device comprising a memory, a processor, the memory storing a computer program, the processor being arranged to perform the method described above by means of the computer program.

In the embodiment of the invention, a similar relation of service resources in a first application is operated according to an account, a second application mode that similar accounts are operated through the first application is pushed to the account, and a first group of target accounts related to a first account in the account set is determined in the account set of the first application, wherein the first service resources operated by the first account in the first application are similar to the second service resources operated by the first group of target accounts in the first application; acquiring a second application set of the first group of target accounts through the first application operation; and pushing part or all of the second application set to the first account, wherein the service resources operated in the first application can describe the interests and hobbies of the user from the side, so that the probability that the first group of target accounts are favored by the user of the first account through the second application operated by the first application and determined based on the similarity relation of the service resources operated in the first application is higher, thereby realizing the technical effect of improving the service resource pushing efficiency and further solving the technical problem of poor pushing efficiency of the service resource pushing method in the related technology.

Drawings

The accompanying drawings, which are included to provide a further understanding of the application and are incorporated in and constitute a part of this specification, illustrate embodiments of the application and together with the description serve to explain the application and do not constitute a limitation on the application. In the drawings:

fig. 1 is a schematic diagram of an application environment of a service resource pushing method according to an embodiment of the present application;

FIG. 2 is a flow chart of an alternative method for pushing service resources according to an embodiment of the present application;

FIG. 3 is a schematic diagram of an alternative service resource pushing method according to an embodiment of the present application;

FIG. 4 is a schematic diagram of another alternative service resource pushing method according to an embodiment of the present application;

FIG. 5 is a flow chart of another alternative method for pushing service resources according to an embodiment of the present application;

FIG. 6 is a flow chart of yet another alternative method of pushing a service resource according to an embodiment of the present application;

FIG. 7 is a schematic diagram of yet another alternative service resource pushing method according to an embodiment of the present application;

FIG. 8 is a schematic diagram of yet another alternative service resource pushing method according to an embodiment of the present application;

FIG. 9 is a flow chart of yet another alternative method for pushing service resources according to an embodiment of the present application;

FIG. 10 is a schematic structural diagram of an alternative service resource pushing device according to an embodiment of the present invention;

FIG. 11 is a schematic structural diagram of another alternative service resource pushing device according to an embodiment of the present invention;

fig. 12 is a schematic structural view of an alternative electronic device according to an embodiment of the present invention.

Detailed Description

In order that those skilled in the art will better understand the present invention, a technical solution in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in which it is apparent that the described embodiments are only some embodiments of the present invention, not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the present invention without making any inventive effort, shall fall within the scope of the present invention.

It should be noted that the terms "first," "second," and the like in the description and the claims of the present invention and the above figures are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged where appropriate such that the embodiments of the invention described herein may be implemented in sequences other than those illustrated or otherwise described herein. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.

In the embodiment of the invention, an embodiment of the service resource pushing method is provided. As an alternative embodiment, the service resource pushing method may be applied, but not limited to, in the application environment shown in fig. 1.

As shown in fig. 1, a plurality of terminals 102 running a first application are connected to a server 106 through a network 104. The first application may log in using an account number. The business resources contained in the first application may be operated in the first application, and one or more second applications may be operated by the first application. The collection of accounts in the first application may include a plurality of accounts for which the terminal 102 is logged into the first application.

Server 106 may obtain information from one or more terminals 102 over network 104 for the traffic resources in the set of accounts that the account operated in the first application. For a first account in the account set, the server 106 may determine a first set of target accounts related to the first account in the account set, where a first service resource operated by the first account in the first application is similar to a second service resource operated by the first set of target accounts in the first application, obtain a second application set operated by the first set of target accounts through the first application, and push one or more second applications to the first account, where the one or more second applications may be part or all of the second application set. The pushing mode can be as follows: application information for one or more applications is sent.

For a terminal that logs in to a first application using a first account, the terminal may, after receiving one or more second applications pushed by the server 106, display the one or more second applications in a target display area of a display interface of the first application.

Alternatively, in this embodiment, the above terminal may include, but is not limited to, at least one of the following (the mobile phone shown in fig. 1 is merely an example): a mobile phone, a tablet computer, a notebook computer, a desktop PC, and a hardware device for running an application program. The network may include, but is not limited to, at least one of: a wired network and a wireless network, the wired network may include, but is not limited to, at least one of: wide area network, metropolitan area network, local area network. The wireless network may include, but is not limited to: bluetooth, WIFI, and other networks that enable wireless communications, such servers may include, but are not limited to: the background server of the first application is a server for pushing the application. The above is merely an example, and the present embodiment is not limited thereto.

According to an aspect of the embodiments of the present invention, a service resource pushing method is provided, where the service resource pushing method may be applied to a server 106 or a terminal 102 shown in fig. 1, and considering occupation of a terminal resource, the service resource pushing method is generally applied to the server 106, as shown in fig. 2, and the method includes:

S202, a first group of target accounts related to a first account in an account set of a first application are determined in the account set of the first application, wherein a first business resource operated by the first account in the first application is similar to a second business resource operated by the first group of target accounts in the first application;

s204, acquiring a second application set of the first group of target accounts through the first application operation;

s206, pushing part or all of the second application set to the first account.

Optionally, in this embodiment, the service resource pushing method may be, but not limited to, applied to a pushing process of a game application, where the service resource pushing method may be, but not limited to, applied to an online video media application, a news application, a social application, and other applications with service resource operation functions, so as to implement pushing a second application interested by a user according to operation service resources. For example, according to the similarity of video resources watched by the user in the online video media application, the game application clicked and downloaded by the user similar to the user through the online video media application is recommended to the user, which is just an example and is not limited in any way in the present embodiment.

The service resource pushing method in the embodiment of the invention is described below with reference to fig. 2.

In step S202, a first set of target accounts related to a first account in the account set is determined in the account set of the first application, where a first service resource operated by the first account in the first application is similar to a second service resource operated by the first set of target accounts in the first application.

The first application may be an online video media application, a news application, a social application, or other application with business resource operations.

The account set may be a set of all accounts in the first application, or may be a set of partial accounts selected from all accounts. The selection method can be as follows: the random extraction from all accounts may be based on the user account system (e.g., whether the third party application account is logged in or registered, and different third party application accounts may belong to different user account systems), may be based on the type of user equipment (e.g., android and iOS), may be based on the gender of the user to which the account corresponds (e.g., male and female), and may also be based on the age of the user to which the account corresponds (e.g., under 10 years old, 10-20 years old, 20-30 years old, 30-40 years old, 40-60 years old, etc.). The above is merely an example, and the present embodiment is not limited thereto.

The first application may provide different types of business resources, e.g., video resources, picture resources, web page resources (text resources), etc. The traffic resources may be represented by resource characteristics. The resource characteristics may be description information of the resource, which may be the resource itself (e.g., web content of a web page), a profile of the resource (e.g., video profile, web summary), or a tag of the resource (e.g., video attribute tag, image attribute tag, web tag).

For example, for a video asset provided by a first application, having a video attribute tag may include: tag 1 (video type, variety, drama, movie, documentary, etc.), tag 2 (video attribute, emotion, reasoning, history, fairy tale, etc.), tag 3 (country of belonged), tag 4 (drama name, e.g., "aaa"), tag 5 (main corner name, e.g., "aaa"), tag 6 (actor name, e.g., "bbb"), etc., and the tags corresponding to the video asset are [ tag 1, tag 2, …, tag 6, … ].

The more the number of tags, the more abundant the information representing the business resources, the higher the accuracy of representing the business resources, and the higher the complexity of computing the similarity. The video attribute tags may be defined as desired, which is not limited in this embodiment.

After the account set is obtained, for a first account in the account set, a first group of target accounts related to the first account can be determined in the account set, wherein a first service resource operated by the first account in the first application is similar to a second service resource operated by the first group of target accounts in the first application. The first business resource may include one or more business resources provided by the first application, and the second business resource may include one or more business resources provided by the first application.

The similarity of the first traffic resource to the second traffic resource may be determined based on the resource characteristics for representing the first traffic resource and the resource characteristics for representing the second traffic resource. The similarity may be semantic similarity of the features (whether the semantics of the features are similar), or term similarity of the features (whether the terms contained in the features are identical).

To simplify the determination that the first traffic resource is similar to the second traffic resource, the user (account number) may be represented vectorially.

As an alternative implementation manner, the vector used for representing the user may be determined according to the label corresponding to the service resource operated by the user in the first application. For a predetermined tag in the tag set, if the predetermined tag is included in a tag corresponding to a service resource operated by a user in a first application, an element value of an element (for example, an element of an nth bit) corresponding to the predetermined tag in a vector representing the user is set to a first predetermined value (for example, 1), and otherwise, is set to a second predetermined value (for example, 0).

For example, if 10 tags, tag 1, tag 2, … …, and tag 10 are included in the tag set, and if tag 1, tag 3, tag 6, and tag 10 are included in the tag corresponding to the service resource operated in the first application, the tag vector corresponding to the user may be represented as [1,0,1,0,0,1,0,0,0,1].

Because different service resources may contain the same tag, the same tag may appear multiple times in the tag corresponding to the service resource operated by the user in the first application. In determining the vector for representing the user, the label appearing a plurality of times may be calculated once or a plurality of times.

For example, if 10 tags, tag 1, tag 2, … …, and tag 10 are included in the tag set and tag 1 (1 time), tag 3 (3 times), tag 6 (1 time), and tag 10 (2 times) are included in the tag corresponding to the service resource operated in the first application, the tag vector corresponding to the user may be represented as [1,0,1,0,0,1,0,0,0,1] or as [1,0,3,0,0,1,0,0,0,2].

As another alternative embodiment, the vector for representing the user (account) may also be determined according to the operation information of the predetermined operation performed on the service resource by the user and the tag corresponding to the service resource.

The operation information of the user performing the operation on the service resource may describe the preference degree of the user on the service from the side, for example, the number of clicking operations, the duration of watching operations, and the like. Thus, the vector representing the user can be determined in combination with the operation information of the predetermined operation that the user has performed on the service resource and the tag corresponding to the service resource.

In an alternative embodiment of the present invention, determining, in the account set of the first application, a first group of target accounts related to a first account in the account set may include: determining a first vector according to first operation information of a first target operation executed by a first account on a first service resource and a first label corresponding to the first service resource; determining a second vector according to second operation information of a first target operation executed by the account in the account set on the second service resource and a second label corresponding to the second service resource; and determining the first group of target accounts from the account set according to the vector similarity between the second vector and the first vector.

The dimensions of the first vector and the second vector can be determined according to the number of different labels corresponding to all the service resources of the first application, or according to the number of different labels corresponding to all the service resources operated in the first application by all the accounts in the account set, or by other methods.

The first target operation may be an interaction behavior of the user with respect to a business resource in the first application, which may include, but is not limited to: clicking on to view a business resource in a first application, the first operational information may include, but is not limited to, at least one of: the time at which the first target operation is performed, the number of times the first target operation is performed, and the like.

In the case where the first operation information of the first target operation includes the number of clicks on each of the first service resources, in order to avoid an influence of a user's false click on the determined tab vector, a viewing time threshold (for example, 5 s) may be set, and no statistics may be made for the click operation whose viewing time is less than the viewing time threshold.

A plurality of first target operations (e.g., multiple clicks) may be performed on the same business resource user, and a tag corresponding to the business resource may be assigned to the user once after each click (or the time period for viewing after the click exceeds a viewing time threshold) by the user.

The same service resource may include multiple independent parts (e.g., a television episode may include different episodes and episodes, a movie may include upper and lower portions, and a variety program may include multiple periods), and labels corresponding to the different parts may be different, so that the multiple independent parts of the same service resource may be used as an individual service resource, and after a user performs a first target operation on the individual service resource (clicks, or a time period for viewing after clicking exceeds a viewing time threshold), the label corresponding to the individual service resource is given to the user once.

The operation of assigning labels to users may be triggered in real time by the user operating the business resources. After the labels of the business resources are assigned to the accounts in the set of accounts, each account may correspond to a set of labels.

There are scenarios where some labels appear on almost all user attribute distributions, and the effect of these labels on user interest characterization is negligible. At the same time, different numbers of tags in the same user have different roles in interest characterization. To more accurately characterize the user's interests, the labels corresponding to the business resources can be regarded as words, the user can be regarded as documents, and the value of each dimension of the label vector can be determined by calculating the TFIDF (term frequency-inverse document frequency) of each label.

Optionally, determining the first vector according to the first operation information of the first target operation executed by the first account on the first service resource and the first label corresponding to the first service resource includes: determining a first tag set corresponding to the first account, wherein the first tag set is obtained by adding a first tag corresponding to a target service resource into the first tag set under the condition that the first account executes a first target operation on the target service resource of the first service resource; determining a first target value corresponding to the target labels in the first label set, wherein the first target value is positively correlated with the quotient of the number of the target labels contained in the first label set and the total number of the labels contained in the first label set, and is positively correlated with the quotient of the total number of the accounts contained in the account set and the number of the accounts contained in the second group of target accounts, and the second label corresponding to the accounts in the second group of target accounts comprises the target labels; and determining a first vector according to the first target value, wherein the element value of the target element corresponding to the target label in the first vector is the first target value.

When determining the first vector, a first set of labels corresponding to the first account is first obtained. And under the condition that the first account executes the first target operation on the target service resource of the first service resource, adding the first label corresponding to the target service resource into the first label set. The first labels corresponding to different resources in the first service resources may be the same or different.

For example, after detecting that the first account performs a clicking operation on the target resource, or performs a clicking operation and the viewing duration exceeds a predetermined time threshold, the first tag corresponding to the target service resource is added to the first tag set, where the first tag set may be directly placed into the first tag set, or may be added according to a tag, if the first tag set already includes the tag, the statistics number of the tag is added by 1, and if the first tag set does not include the tag, the statistics number of the tag is added to the tag, and is set to 1.

After the first set of labels is obtained, the values of the dimensions (element values of the elements) in the first vector are calculated.

For a target tag in the first set of tags, the value of the dimension in the first vector corresponding to the target tag (i.e., the first target value) may be determined from two values: in the first set of labels, the number of target labels is proportional to the total number of labels in the first set of labels (first proportion, positive correlation), and the number of accounts containing target labels in the corresponding set of labels is proportional to the total number of accounts in the set of accounts (second proportion, negative correlation). The correlation function and the correlation coefficient between the first target value and the first proportion value, and the correlation function and the correlation coefficient between the first target value and the second proportion value can be set according to the requirement.

For example, the labels corresponding to the business resources can be regarded as words, the user can be regarded as documents, the word frequency-inverse document frequency corresponding to each label in the first label set is determined, the calculation formula is shown in formula (1),

TFIDF _i,j ＝tf _i,j *idf _i (1)

wherein, the liquid crystal display device comprises a liquid crystal display device,n _i,j indicating that tag ti is on user d _j In number of occurrences, |D| represents the total number of all users, | { j: t ε D) _j The sign t appears in the document indicating all users _i Is a number of documents. To reduce the complexity of the similarity progression, labels with idf values less than a given threshold may be filtered out.

After obtaining the values (e.g., TFIDF values) of the dimensions corresponding to the labels in the first label set, the first account may be represented as a first vector, where the dimension size of the vector is the total number of labels, and the value of each dimension of the vector is the TFIDF value of the corresponding label. For a tag that is not included in the first set of tags, the dimension corresponding to the tag may be set to 0.

It should be noted that, the manner of determining the first vector is only shown above, and the manner of determining the second vector corresponding to other accounts in the account set is similar to the first vector, which is not described herein.

After the vectorized representation of each account is obtained (the first vector and the second vector), a vector similarity between each second vector and the first vector may be determined, and a vector similarity between any two second vectors may also be determined. The vector similarity may be determined by adopting a cosine similarity calculation method, or may be determined by adopting other methods for calculating the vector similarity, which is not particularly limited in this embodiment.

After obtaining the vector similarity between each second vector and the first vector, the first group of target accounts can be determined from the account set according to the vector similarity between the second vector and the first vector.

As an alternative embodiment, the manner of determining the first set of target accounts may be: and selecting a preset number (M) of target accounts from the account set as the first group of target accounts according to the magnitude of the vector similarity between the second vector and the first vector. The manner in which the target account number is selected may include a variety of ways.

The selection method can be as follows: sequentially selecting accounts from the account set according to a set threshold (preset threshold), taking the selected accounts as target accounts in the first group of target accounts if the similarity of the second vector used for representing the selected accounts and the vector of the first vector is larger than the preset threshold, otherwise, continuing to select the accounts; until the number of 5 target accounts in the first group of target accounts reaches a predetermined number.

The selection mode can also be as follows: and ordering the accounts in the account set according to the sequence from large to small of the vector similarity between the second vector and the first vector corresponding to each target account in the account set, and selecting the accounts with the previous preset number as the first group of target accounts.

As another alternative embodiment, the manner of determining the first group of target accounts may be: and determining a first group of target accounts in the account set at least according to the vector similarity between a third vector and the first vector and the vector similarity between a fourth vector and the third vector, wherein the third vector is a vector corresponding to a second account with a first-order neighbor relation with the first account in a target relation network, the fourth vector is a third account with a second-order neighbor relation with the first account through the second account in the target relation network, the target relation network is a non-directional weighted relation network, the nodes of the target relation network represent accounts in the account set, and the weight on the edge between two nodes in the target relation network is the vector similarity between the second vectors corresponding to the two nodes.

The user's interaction with the second application (e.g., browsing, clicking or downloading) is not affected by the user's first-order neighbors. That is, the local influence of the user may affect not only the first-order neighbors but also the second-order neighbors and even the multi-order neighbors. For example, a friend of a user's friend may have some impact on the user's preference for a second application (e.g., game).

After determining the similarity between each second vector and the first vector, and the similarity between any two second vectors, the calculated similarity (i.e., the relationship between the users) may be mapped onto a target relationship network to form an undirected weighted user relationship network G, i.e., w→g (U, E, W), where U is the set of users and E is the set of edges between the users, and in the user relationship network, the nodes represent the users (accounts). W (W) _i,j Is shown for representing user u _i Vector sum for representing u _j If W _i,j >0, indicating u in the user relationship network _i And u _j Between which there is an edge E _i,j And the weight on the edge is W _i,j . After mapping the user relationships to the network, as shown in FIG. 3, more relationships between users can be explored using the method of graph computation.

In the user relationship network, nodes represent users, and weights on edges represent weights of relationships between users. The theory of analog information propagation, the user has a certain influence on the decision of its first-order or multi-order neighbors. For example, having message x, which is propagated by user a to friend B of a, which in turn propagates the message x to friend C of B, user C may make a decision regarding x when hearing the message x.

After mapping the relationships between users onto the graph to obtain a network of user relationships, local or k-order neighbor relationships of the users can be inferred. As shown in fig. 3, fig. 3 shows a first-order neighbor and a second-order neighbor of node a. The first group of target accounts may be determined in the account set based at least on a similarity of the first account to its first-order neighbors and a similarity of the first-order neighbors to the second-order neighbors.

For a scenario in which a first group of target accounts is determined according to a first-order neighbor and a second-order neighbor, the determination manner of the first group of target accounts may be: determining the similarity of the first account and the first-order neighbor; determining the similarity of the first account and the second-order neighbor; and determining the first group of target accounts from the first-order neighbors and the second-order neighbors according to the size of the similarity.

The similarity of the first account number and its second order neighbors may be: a weighted product of the first similarity and the second similarity. The first similarity is the similarity between the first account and its first-order neighbor (the node between the first account and its second-order neighbor), and the second similarity is the similarity between its first-order neighbor and its second-order neighbor. The weighted weight may be empirically set in consideration of the attenuation of the influence in the indirect association.

For a scene with a plurality of first-order neighbors between the first account and the second-order neighbors, the similarity of the first account and the second-order neighbors corresponding to the first-order neighbors can be calculated respectively, and then the obtained similarity is summed.

For example, node b is a second-order neighbor of node a, and is connected by node c and node d, respectively, and the similarity between node b and node a can be expressed as: f1×W _a,c *W _c,b +f2*W _a,d *W _d,b Wherein f1 and f2 are weighting coefficients, and may be the same or different.

For the nodes higher than the second order, the similarity is calculated in a similar manner to the second order, and the difference is mainly different from the weighting coefficient, which is not described herein.

It should be noted that if there are a plurality of relationships (for example, both a first-order neighbor and a second-order neighbor) between two nodes, only the nearest relationship is selected as the relationship between the two nodes. As shown in fig. 3, the node b2 may be considered as a first-order neighbor of the node a, and may be considered as a second-order neighbor of the node a (via the node b 1), and the relationship between the two is only a first-order neighbor relationship.

Optionally, in the embodiment of the present invention, a "sampling and aggregating" strategy is used to perform deep random walk on the user relationship network, so as to mine the relationship between users in a deeper level: on the one hand, the huge number of nodes in the user relation network leads to the time and effort consumption of directly solving the local or even global influence of the user; on the other hand, it is not possible for the user to conduct message propagation with all of his first-order or even multi-order friends, e.g., user A has 100 first-order friends, but A remains in daily contact with only 10 of his friends.

The sampling strategy refers to: according to the current user node, sampling a first-order neighbor node of the current user node according to a given strategy; the aggregation policy refers to that the sampled information of k-order users is aggregated according to a given policy so as to serve the current user node.

The generation mode of the depth random walk sequence is that a node which is currently walked is assumed to be vi, a neighbor node is randomly sampled from a first-order neighbor of the node vi according to weight, and the neighbor node is used as a current node which is reached by the random walk sequence. Each random walk is stopped under conditions that reach a set threshold length.

Assuming S as a database for model training, S is obtained by gathering all random walk sequences, n ^s A is the walk length of the random walk sequence s _j ＝{v _j-c ,…,v _j+c Is given as target node v _j Context nodes of (a), aggregating information of the context nodes as node v _j And (5) serving. The goal of the model is to: the low-dimensional vectorization of each node (user) in order to maximize the following objective function:

wherein the probability P (v _i |v _j ) The value of (2) is entered using a softmax functionRow calculation:

wherein o is _m And o' _m The input and output vectorized representations of node m, respectively. To speed up the time of model training, the objective function described above may be solved approximately using the Hierarchical Softmax method, resulting in a low-dimensional representation vector for each user. Specifically, as shown in fig. 4, each user node is regarded as a leaf node of a binary tree, and P (v _i |v _j ) Is transformed into the probability size of a particular path in the hierarchical tree. The assumed paths are uniquely determined by nodes in a series of trees, f ₀ ,f ₁ ,…,f _m Then P (v) _i |v _j ) The calculation formula of (2) is shown as the formula:

wherein, the liquid crystal display device comprises a liquid crystal display device,representing a binary tree node f _t Is a vectorized representation of the parent node of (c). By the above-mentioned process, P (v _i |v _j ) Is reduced from O (|u|) to O (log|u|).

For the depth random walk strategy, random sampling can be performed according to the weight of the edge, and the probability that the node returns to the node through two steps of walk is reduced; other walk strategies may also be employed, such as limiting the size of the probability of a walk from a particular node, the randomness of the walk length, etc.

The random walk strategy described above assumes a strong relationship between users within a window on the same walk sequence. Other aggregation strategies may also be employed, such as explicitly defining k aggregation functions, collecting k-th order neighbor information for the target user, adjustment of the wander window parameters, and so forth.

The random walk strategy described above employs a random walk strategy that is unsupervised with respect to the final second application recommendation target, and may also employ a supervised random walk strategy.

After the low-dimensional representation vector of each account is obtained, the first group of target accounts can be determined from the account set according to the similarity between the low-dimensional representation vector of the accounts except the first account in the account set and the low-dimensional representation vector of the first account. The determination is similar to the foregoing, and will not be described in detail herein.

In step S204, a second application set of the first group of target accounts operated by the first application is acquired.

In a first application, pushing of a plurality of second applications may be provided. The user may perform an operation (e.g., click, download, etc.) on the second application according to the provided push. For the first group of target accounts, a second application operated by each target account through the first application can be obtained, and a second application set is obtained.

For example, the second application is a game, the first account is user a, and the first set of target accounts includes: the game operated by the user B through the first application comprises the following steps: game g ₁ Game g ₂ Game g ₃ The game operated by the user B through the first application comprises the following steps: game g ₁ Game g ₂ Game g ₃ And game g ₄ The second set of applications includes: { g ₁ ,g ₂ ,g ₃ ,g ₄ }。

In step S206, part or all of the second application set is pushed to the first account.

After the second application set is obtained, all or part of the second application set may be pushed to the first account. All of the second application set may be pushed to the first account or a randomly selected portion of the second application set may be pushed to the first account.

In order to improve the interest degree of the user in the pushed second application and improve the pushing efficiency, before pushing part or all of the second application set to the first account, the second applications in the second application set may be first ordered.

Optionally, the second applications in the second application set are ordered according to third operation information of the second target operation executed by the second application in the second application set by the first application according to each target account in the first group of target accounts.

When the second applications in the second application set are ordered, firstly, according to each target account in the first group of target accounts, determining a recommendation index of each second application through third operation information of a second target operation executed by the first application on the second applications in the second application set; the second applications are ranked according to the recommendation index for each second application.

The recommendation index for each second application may be determined in a number of ways.

As an alternative embodiment, the number of accounts in the first set of target accounts that operate the second application through the first application may be used as the recommendation index of the second application: the second applications may be ordered in order of more than less number of accounts in the first set of target accounts that operate the second applications through the first application.

As another alternative embodiment, the result of summing the similarity between the target account and the first account, which have been operated by the first application, of the second application may be taken as the recommendation index of the second application: summing the similarity between the target account operated by the first application and the first account to obtain a recommendation index of the second application; the second applications are ordered according to the order of the recommendation index from high to low.

As yet another alternative embodiment, the recommendation index of each second application may be determined according to the preference value of each target account for the second application and the similarity between each target account and the first account; the second applications are ordered according to the order of the recommendation index from high to low.

Optionally, sorting the second applications in the second application set according to the third operation information of the second target operation executed by the second application in the second application set by the first application by each target account in the first group of target accounts includes: determining a preference value of each target account for the second application according to the operation times of the second application executed by each target account for the second application in the second application set through the first application, wherein the preference value is a result obtained by carrying out weighted summation on the operation times through target weights; according to the vector similarity between the fifth vector used for representing each target account and the sixth vector used for representing the first account, weighting and summing the preference values of the second application by each target account to obtain a second target value corresponding to the second application; and sequencing the second applications in the second application set according to the order of the second target values corresponding to the second applications in the second application set from the large to the small.

The second target operation performed by the first application on the second application in the second application set may be various, for example, clicking, downloading, opening, specifically, clicking (not installed), downloading completion, registering completion, downloading pause, downloading resume, downloading cancel, downloading failure, opening (installed). The third operation information of the second target operation may be: the number of operations of each second target operation is performed.

When determining the preference value of the target account for the second application, an empirical weight may be given to the second target operation, and the preference value of the target account for the second application is: a weighted summation of the number of operations of each second target operation is performed.

The manner in which the preference value of the target account for the second application is determined will be described below in connection with specific examples. In the scenario that the second application is a game, the user clicks and downloads the interactive behavior of the game (the interactive behavior is uniquely determined by time, user account system type, user id and game id), and the method specifically includes clicking, downloading completion, registering completion, downloading pause, downloading recovery, downloading cancellation and downloading failure. The above-described decomposed behaviors may be respectively given empirical weights 0.5,0.6,0.7,0.8,0.3,0.8,0.2,0.5. Meanwhile, the clicking and downloading times of each user can be obtained. Solving user preference values for games by adopting weighted summation mode r, i.e. r _u,i ＝x ₁ y ₁ +x ₂ y ₂ +…+x _K y _K Wherein r is _u,i For user u to play the preference value of game i, x _k Representing the game behavior weight of the user, y _k Representing the number of corresponding game actions, K representing the number of game actions.

After obtaining the preference value of each target account for the second application, the preference value of each target account for the second application can be weighted and summed according to the similarity between each target account and the first account, so as to obtain the recommendation index of the second application.

After the second applications are ordered according to the order from high to low of the recommendation index, a candidate second application list can be obtained, all second applications in the second application list can be sequentially pushed to the first account, and the first n second applications in the second application list can be sequentially pushed to the first account.

The manner in which the candidate second application list is determined is described below in connection with specific examples. In this example, the second application is a game.

After the low-dimensional vectorization representation of each user under the same vector space is obtained, for a given target user, calculating weights of other users and the target user by adopting a cosine similarity calculation method, sorting the weights, and taking m corresponding users before sorting as a user set U with similar target users.

After obtaining the preference value of each user to each game, the candidate game list can be recommended to the target user (the user corresponding to the first account) by utilizing the thought of collaborative filtering based on the user, and the calculation formula of the recommendation index is shown in formula (4):

wherein r is _u,i Representing the preference value of user u for game i, W _u,u’ The weight between the users u and u' is represented, and the function aggr adopts a weighted summation modeThe game preference values are integrated. And obtaining a candidate game recommendation list of the target user by sequencing the preference value of the user to the corresponding game.

For example, the target user is a, the users similar to u are user B and user C, the corresponding weights are 0.8 and 0.6, and the game behavior preferences of users A, B and C are respectively: { g ₁ :0.5}，{g ₁ :0.5,g ₂ :0.8,g ₃ :0.4}，{g ₁ :0.2,g ₂ :0.3,g ₃ :0.8,g ₄ 0.9, then target user A is for game g ₂ Preference values of 0.8 x 0.8+0.6 x 0.3=0.82, and the same applies to a for g ₃ And g ₄ The preference values of (2) are 0.8 and 0.54, and thus the game recommendation list of the target user A is { g } ₂ :0.82,g ₃ :0.8,g ₄ :0.54}。

In this embodiment, a manner of pushing a second application operated by a first application to an account by using a similar relationship between service resources in a first application operated by an account is adopted, and a first group of target accounts related to a first account in the account set is determined in the account set of the first application, where a first service resource operated by the first account in the first application is similar to a second service resource operated by the first group of target accounts in the first application; acquiring a second application set of the first group of target accounts through the first application operation; part or all of the second application set is pushed to the first account, the technical problem that the pushing efficiency is poor in the service resource pushing method in the related technology is solved, and the service resource pushing efficiency is improved.

As an optional technical solution, determining, in an account set of the first application, a first group of target accounts related to a first account in the account set includes:

s1, determining a first vector according to first operation information of a first target operation executed by a first account on a first service resource and a first label corresponding to the first service resource;

s2, determining a second vector according to second operation information of a first target operation executed by the account in the account set on the second service resource and a second label corresponding to the second service resource;

s3, determining a first group of target accounts from the account set according to the vector similarity between the second vector and the first vector.

According to the embodiment of the invention, the users are represented in a vectorization manner, and the similarity between the users (the relationship between the users) is represented based on the similarity between the vectors, so that the determination process of the similarity between the users is simplified, and the determination process of the first group of target accounts is improved.

Optionally, determining the first vector according to the first operation information of the first target operation executed by the first account on the first service resource and the first label corresponding to the first service resource includes:

S11, determining a first tag set corresponding to the first account, wherein the first tag set is obtained by adding a first tag corresponding to a target service resource to the first tag set under the condition that the first account executes a first target operation on the target service resource of the first service resource;

s12, determining a first target value corresponding to the target labels in the first label set, wherein the first target value is positively correlated with the quotient of the number of the target labels contained in the first label set and the total number of the labels contained in the first label set, and is positively correlated with the quotient of the total number of the accounts contained in the account set and the number of the accounts contained in the second group of target accounts, and the second labels corresponding to the accounts in the second group of target accounts comprise the target labels;

s13, determining a first vector according to the first target value, wherein the element value of the target element corresponding to the target label in the first vector is the first target value.

According to the embodiment of the invention, the capability of the label for representing the user is integrated when the vector for representing the user is determined, so that the accuracy of the vector for representing the user is improved.

Optionally, determining the first group of target accounts from the account set according to the vector similarity between the second vector and the first vector includes:

S31, determining a first group of target accounts in the account set, wherein the vector similarity between a second vector corresponding to each target account in the first group of target accounts and the first vector is larger than a preset threshold; or alternatively, the process may be performed,

s32, determining the first M target accounts in the account set as a first group of target accounts according to the sequence of the vector similarity between the second vector corresponding to each target account in the account set and the first vector from large to small, wherein M is a natural number.

According to the embodiment of the invention, the first group of target accounts are determined from the account set in a plurality of modes, so that the method and the device can be suitable for different scenes, and the diversity of the determination modes of the first group of target accounts is improved.

As an optional technical solution, determining the first group of target accounts from the account set according to the vector similarity between the second vector and the first vector includes:

s33, determining a first group of target accounts in the account set at least according to the vector similarity between a third vector and the first vector and the vector similarity between a fourth vector and the third vector, wherein the third vector is a vector corresponding to a second account with a first-order neighbor relation with the first account in a target relationship network, the fourth vector is a third account with a second-order neighbor relation with the first account through the second account in the target relationship network, the target relationship network is an undirected weighted relationship network, the nodes of the target relationship network represent accounts in the account set, and the weight on the edge between two nodes in the target relationship network is the vector similarity between the second vectors corresponding to the two nodes.

By the embodiment of the invention, the first group of target accounts is determined at least according to the second-order neighbor relation of the user, so that the accuracy of determining the first group of target accounts can be improved.

As an optional technical solution, pushing part or all of the second application set to the first account includes:

s1, sorting second applications in a second application set according to third operation information of second target operations executed by each target account in a first group of target accounts through the second applications in the second application set;

s2, selecting part or all of the second applications from the second application set according to the sorting result obtained by sorting, and pushing the second applications to the first account.

By the embodiment of the invention, the second application can be pushed according to the possible interested degree of the user by sequencing and pushing the applications in the second application set, so that the pushing efficiency is improved.

As an optional technical solution, according to third operation information of a second target operation performed by each target account in the first group of target accounts on a second application in the second application set through the first application, ordering the second application in the second application set includes:

S11, determining a preference value of each target account for the second application according to the operation times of the second target operation executed by the second application in the second application set by the first application, wherein the preference value is a result obtained by carrying out weighted summation on the operation times by the target weight;

s12, according to the vector similarity between the fifth vector used for representing each target account and the sixth vector used for representing the first account, weighting and summing the preference values of the second application by each target account to obtain a second target value corresponding to the second application;

and sequencing the second applications in the second application set according to the order of the second target values corresponding to the second applications in the second application set from the large to the small.

According to the embodiment of the invention, the similarity between the users and the preference degree of the users for the second application are comprehensively considered for the second application, and the accuracy of the sequencing result is improved.

As an alternative to this,

determining, in an account set of a first application, a first set of target accounts related to a first account in the account set includes: determining a first group of target accounts related to a first account in the account set of the first application, wherein a first video resource operated by the first account in the first application is similar to a second video resource operated by the first group of target accounts in the first application;

The method for acquiring the second application set of the first group of target accounts operated by the first application comprises the following steps: and acquiring a game application set of the first group of target accounts operated by the first application.

According to the embodiment of the invention, the second application operated by the similar user through the first application is recommended to the user according to the similarity of the operation video resources in the first application of the user, so that the possibility of recommending the game by the user operation can be improved, and the efficiency of game pushing is improved.

The following describes a service resource pushing method in this embodiment in combination with a specific example. In this example, the business resource provided by the first application is a video resource and the second application is a game. FIG. 5 illustrates a depth random walk-based game recommendation algorithm based on which games of possible interest to a user may be recommended.

According to another aspect of the embodiment of the present invention, there is further provided a service resource pushing method, which may be applied to the terminal 102 shown in fig. 1. As shown in fig. 6, the method includes:

s602, determining that a display page of a client of a first application logged in by using a first account is slid to a target display area, wherein the target display area is used for displaying one or more second applications pushed to the first account, the one or more applications are part or all of a second application set, the second application set is a set of second applications, which are operated by a first group of target accounts through the first application, the first group of target accounts are a group of accounts related to the first account, which are determined from an account set of the first application, and first business resources operated by the first account in the first application are similar to second business resources operated by the first group of target accounts in the first application;

And S604, pushing one or more second applications to the first account in the target display area.

The service resource pushing method in the embodiment of the present invention is described below with reference to fig. 6.

In step S602, it is determined that the display page of the client of the first application registered using the first account is slid to the target display area.

The client operates in a user terminal (e.g., terminal 102) that logs in using a first account number. The target display area may be located in a current display interface (e.g., a main interface, a display interface of a business resource) of the client. The target display area may be located at a predetermined position of the current display interface, and may be slid to the target display area by sliding the current display interface up and down, or the target display area may be located on an independent display interface or an independent display area, for example, by switching the display interface, the display interface of the client is slid to the target display area, or by calling the target display area through a sliding operation in the current display interface, the called target display area may be located at an upper layer of the current display interface, or may be added to the current display interface.

After step S602, the client may send a request message to the server through the user terminal, where the request message is used to request to obtain application information of one or more second applications pushed to the first account; and receiving a response message sent by the server and used for responding to the request message, wherein the response message carries application information of one or more second applications.

The application information of the one or more second applications may include, but is not limited to: an application icon of the second application, a link for jumping to a download interface of the second application, and the like.

The manner in which the server determines the one or more second applications pushed to the first account in response to the request message is similar to that described above, and will not be described in detail herein.

In step S604, one or more second applications pushed to the first account in the target display area.

After receiving the one or more second applications pushed by the server, the one or more second applications may be displayed in the target display area. There are various ways of displaying one or more second applications in the target display area, for example, a first number of second applications may be displayed in the target display area, and after detecting a sliding operation or clicking operation performed by the user on the target area, the remaining second applications are sequentially displayed, or a switch is made to other first number of second applications, or the like. For another example, a first number of second applications may be displayed in the target display area, and after detecting a touch operation on the target button, a jump may be made to a display interface of one or more second applications. The above is merely an example, and the present embodiment is not limited thereto.

For example, as shown in fig. 7, the service resource provided by the first application is a video resource, the second application is a game, and when the video display interface is slid down to a display area for displaying the pushed game, the pushed game is displayed in the display area. Since the user does not install the pushed game, the "download" button of the game is displayed under the game icon. As another example, as shown in fig. 8, since the user has installed the pushed game, the "enter" button of the game is displayed under the game icon.

Optionally, after step S604, detecting a target operation performed on a target application of the one or more second applications; under the condition that a target application is installed at a terminal where the client is located, responding to target operation, and jumping to the target application; or, if the terminal where the client is located is not provided with the target application, responding to the target operation and jumping to a downloading interface of the target application.

The user may operate the second application pushed by the target display area, for example, perform a target operation (click, double click, slide), jump to the target application (for example, start the target application) if the operated second application is installed on the user terminal, jump to a download interface of the operated second application if the operated second application is not installed on the user terminal.

In this embodiment, a manner of pushing a second application operated by a first application to an account by using a similar relationship between service resources in a first application operated by an account is adopted, and a first group of target accounts related to a first account in the account set is determined in the account set of the first application, where a first service resource operated by the first account in the first application is similar to a second service resource operated by the first group of target accounts in the first application; acquiring a second application set of the first group of target accounts through the first application operation; when the display page of the client is detected to be slid down to the target display area, one or more second applications pushed to the first account are displayed in the target display area, so that the technical problem that the pushing efficiency is poor in the service resource pushing method in the related art is solved, and the service resource pushing efficiency is improved.

As an optional technical solution, after determining that the display page of the client of the first application logged in using the first account is slid down to the target display area, the method further includes:

S1, sending a request message to a server, wherein the request message is used for requesting to acquire application information of one or more second applications pushed to a first account;

s2, receiving a response message sent by the server and used for responding to the request message, wherein the response message carries application information of one or more second applications.

According to the embodiment of the invention, after the display page of the client is slid down to the target display area, the application information of the pushed one or more second applications is obtained in a request-response mode, so that the instantaneity of the pushed one or more second applications can be ensured, and the accuracy of service resource pushing is improved.

As an optional technical solution, after the target display area pushes one or more second applications to the first account, the method further includes:

detecting a target operation performed on a target application of the one or more second applications;

under the condition that a target application is installed at a terminal where the client is located, responding to target operation, and jumping to the target application; or alternatively, the process may be performed,

and under the condition that the terminal where the client is located is not provided with the target application, responding to the target operation and jumping to a downloading interface of the target application.

According to the embodiment of the invention, after the target operation executed on the target application is detected, the target application is jumped to or the download interface of the target application is jumped to, so that the user can conveniently operate the pushed second application, and the user experience is improved.

The service resource pushing method is described below with reference to specific examples. In this example, the service resources provided by the first application include: the video resource and the second application is a game.

The business resource pushing method provided in the example constructs a relationship network between users through video labels, deduces a local or k-order neighbor relationship of the users on the relationship network, and then based on the deduced depth relationship and the known interaction data of the users and the games, realizes personalized game recommendation by utilizing ideas representing learning and user collaborative filtering.

Methods of collaborative filtering may include, but are not limited to: memory-based methods and model-based methods. The main idea of the method is to calculate the relationship between users according to the interaction behavior of different users on games, preference data or self attributes, so as to recommend games liked by users similar to the user.

The business resource pushing method in the example is designed aiming at the following contents:

1) In a user relationship network, the interactive behavior of a user to a game, such as browsing, clicking or downloading, is not affected by the first-order neighbors of the user. That is, the local influence of the user may affect not only the first-order neighbors but also the second-order neighbors and even the multi-order neighbors. For example, friends of friends may have some impact on our gaming preferences.

2) If the daily life user data of the first application is huge, the calculation complexity of inputting the whole user relation network as a model is very high, and the model needs a great amount of time to train.

The service resource pushing method in the example mainly discovers the implicit interest of the user according to the behavior records of the user, such as the watched video, the interaction behavior with the game and the like, so as to conduct personalized game recommendation of the user.

As shown in fig. 9, the service resource pushing method in this example includes the following steps:

s1, constructing a user relation network.

The server may periodically (e.g., one day, two days, etc.) obtain a list of candidate game recommendations for each user. To obtain a list of candidate game recommendations for each user, a user relationship network may be first constructed.

Because only a small number of users have direct interaction behaviors on the game, a user relationship network can be constructed according to other behaviors of the users, so that the relationship among the users is searched. When the user uses the first application, the user can watch the interested video according to the interest of the user. Thus, when a user views a video, the video tag corresponding to the video, such as "name of a television show", "category" and the like, is regarded as the video attribute tag of the user. Thus, each user has one or more video attribute tags.

Regarding the video attribute tags as words, regarding users as documents, calculating the TFIDF value of each tag of each user, and filtering out tags with idf values smaller than a given threshold, wherein each user can be represented as a video attribute tag vector, the dimension size of the vector is the total amount of the tags, and the value of each dimension of the vector is the TFIDF value of the corresponding tag.

According to the obtained user video attribute label vector, we calculate the relationship between users by adopting a simple and direct cosine similarity method, and map the relationship onto a graph to form an undirected weighted user relationship network G, namely W-G (U, E, W), wherein W _i,j ＝cosine(v _i ,v _j )，v _i And v _j For user u _i And u _j U is a user set, E is a set of edges between users; if W is _i,j >0, indicating u in the user relationship network _i And u _j Between which there is an edge E _i,j And the weight on the edge is W _i,j . The constructed user relationship network may be as shown in fig. 3.

S2, according to a depth random walk strategy, each node is represented in a low-dimensional vectorization mode.

The manner in which the low-dimensional vectorization represents each node (user) according to the deep random walk strategy is similar to that described above, and will not be described in detail here.

S3, determining preference values of the users for games.

The manner of calculating the preference value is similar to that described above and will not be described in detail here.

S4, acquiring a candidate game recommendation list.

After each node is represented by a low-dimensional vectorization, a low-dimensional vectorization representation of each user can be obtained under the same vector space. For a given target user (first account), calculating weights of other users and the target user by adopting a cosine similarity calculation method, sorting the weights, and taking m corresponding users before sorting as a user set U which is similar to the target user.

According to the similarity between each user and the target user in the user set U and the preference value of each user for the game, a candidate game recommendation list is obtained, and the obtaining manner is similar to the foregoing, and is not repeated here.

S5, pushing the obtained candidate game recommendation list to the client.

After acquiring the candidate game recommendation list of each user, the server pushes the acquired candidate game recommendation list to each user (i.e., the client on which each user logs).

S6, detecting that the video page is pulled down to the game horizontal position.

The user pulls down the video page to the game stumbling area (game stumbling) while watching the video using the first application. The game pushed by the server can be displayed at the game sideslip position, and the user can browse, click or download the interesting game.

S7, displaying the games in the candidate game recommendation list at the game horizontal line.

When the video page is pulled down to the game cross-bar, games in the candidate game recommendation list pushed by the server may be displayed at the game cross-bar.

It should be noted that, the candidate game recommendation list may also be actively requested from the server by the client in a request-response manner. The server periodically generates a candidate game recommendation list for each user (periodically updates each user to generate a candidate game recommendation list). And the client sends a request message to the server to request the candidate game recommendation list of the user when detecting that the user opens the video page or detects that the video page is pulled down to the game sideslip place.

The candidate game recommendation list may be a candidate list for a game list at the final game cross-section. The generation of the final game list requires a combination of factors such as the combination of candidate lists generated by multiple algorithms, redefinition of the game recommendation list by the operator, manual intervention, etc.

The click through rate CTR (Click Through Rate) may be used as an evaluation index, and the method may be used to push the service resource in this example.

By way of this example, a "sample and gather" strategy is employed to compute the local or k-th order neighbor relation of the user and to vectorize the user using the deep random walk representation learning concept. The solved user vector is not only used for game recommendation, but also used for algorithm input of other machine learning applications.

It should be noted that, for simplicity of description, the foregoing method embodiments are all described as a series of acts, but it should be understood by those skilled in the art that the present invention is not limited by the order of acts described, as some steps may be performed in other orders or concurrently in accordance with the present invention. Further, those skilled in the art will also appreciate that the embodiments described in the specification are all preferred embodiments, and that the acts and modules referred to are not necessarily required for the present invention.

According to still another aspect of the embodiment of the present invention, there is further provided a service resource pushing device for implementing the service resource pushing method. As shown in fig. 10, the apparatus includes:

(1) A first determining unit 1002, configured to determine, in an account set of a first application, a first group of target accounts related to a first account in the account set, where a first service resource operated by the first account in the first application is similar to a second service resource operated by the first group of target accounts in the first application;

(2) An obtaining unit 1004, configured to obtain a second application set of the first group of target accounts through the first application operation;

(3) A pushing unit 1006, configured to push part or all of the second application set to the first account.

Optionally, in this embodiment, the service resource pushing device may be located in a server, and may be, but not limited to, applied in a pushing process of a game application, where the service resource pushing method may be, but not limited to, applied in an online video media application, a news application, a social application, and other applications with service resource operation functions, so as to implement pushing, according to an operation service resource, a second application interested by a user. For example, according to the similarity of video resources watched by the user in the online video media application, the game application clicked and downloaded by the user similar to the user through the online video media application is recommended to the user, which is just an example and is not limited in any way in the present embodiment.

Alternatively, in the present embodiment, the first determining unit 1002 may be configured to perform the foregoing step S202, the obtaining unit 1004 may be configured to perform the foregoing step S204, and the pushing unit 1006 may be configured to perform the foregoing step S206.

As an alternative solution, the first determining unit 1002 includes:

(1) The first determining module is used for determining a first vector according to first operation information of a first target operation executed by the first account on the first service resource and a first label corresponding to the first service resource;

(2) The second determining module is used for determining a second vector according to second operation information of a first target operation executed by the account in the account set on the second service resource and a second label corresponding to the second service resource;

(3) And the third determining module is used for determining the first group of target accounts from the account set according to the vector similarity between the second vector and the first vector.

Optionally, the first determining module includes:

(1) The first determining submodule is used for determining a first tag set corresponding to the first account, wherein the first tag set is obtained by adding a first tag corresponding to a target service resource into the first tag set under the condition that the first account executes a first target operation on the target service resource of the first service resource;

(2) A second determining submodule, configured to determine a first target value corresponding to a target label in the first label set, where the first target value is positively correlated with a quotient of a number of target labels included in the first label set and a total number of labels included in the first label set, is positively correlated with a quotient of a total number of accounts included in the account set and a number of accounts included in the second group of target accounts, and where a second label corresponding to an account in the second group of target accounts includes the target label;

(3) And the third determining submodule is used for determining a first vector according to the first target value, wherein the element value of the target element corresponding to the target label in the first vector is the first target value.

Optionally, the third determining module includes:

(1) A fourth determining submodule, configured to determine a first group of target accounts in the account set, where a vector similarity between a second vector corresponding to each target account in the first group of target accounts and the first vector is greater than a predetermined threshold; or alternatively, the process may be performed,

(2) And a fifth determining submodule, configured to determine, according to an order of from high to low, a vector similarity between a second vector corresponding to each target account in the account set and the first vector, a first M target accounts in the account set as a first group of target accounts, where M is a natural number.

As an optional solution, the third determining module includes:

(1) A sixth determining submodule, configured to determine a first group of target accounts in the account set according to at least a vector similarity between a third vector and the first vector and a vector similarity between a fourth vector and the third vector, where the third vector is a vector corresponding to a second account having a first order neighbor relation with the first account in the target relationship network, the fourth vector is a third account having a second order neighbor relation with the first account through the second account in the target relationship network, the target relationship network is an undirected weighted relationship network, a node of the target relationship network represents an account in the account set, and a weight on an edge between two nodes in the target relationship network is a vector similarity between the second vectors corresponding to the two nodes.

As an optional solution, the pushing unit 1006 includes:

(1) The ordering module is used for ordering the second applications in the second application set according to third operation information of the second target operation executed by the second applications in the second application set through the first applications by each target account in the first group of target accounts;

(2) And the selecting module is used for selecting part or all of the second applications from the second application set according to the sorting result obtained by sorting and pushing the second applications to the first account.

As an optional technical solution, the sorting module includes:

(1) A seventh determining submodule, configured to determine, according to the operation times of the second target operation performed by the second application in the second application set by the first application by each target account, a preference value of the second application by each target account, where the preference value is a result obtained by performing weighted summation on the operation times by the target weights;

(2) The summation sub-module is used for carrying out weighted summation on the preference value of the second application by each target account according to the vector similarity between the fifth vector used for representing each target account and the sixth vector used for representing the first account, so as to obtain a second target value corresponding to the second application;

(3) And the sequencing sub-module is used for sequencing the second applications in the second application set according to the sequence from the big to the small of the second target values corresponding to the second applications in the second application set.

As an alternative solution, the first determining unit 1002 includes: the third determining module, the acquiring unit 1004 includes: an acquisition module, wherein,

(1) A third determining module, configured to determine, in an account set of a first application, a first group of target accounts related to a first account in the account set, where a first video resource operated by the first account in the first application is similar to a second video resource operated by the first group of target accounts in the first application;

(2) The acquisition module acquires a game application set operated by a first group of target accounts through a first application.

According to still another aspect of the embodiment of the present invention, there is further provided a service resource pushing device for implementing the above service resource pushing method, as shown in fig. 11, where the device includes:

(1) A second determining unit 1102, configured to determine that a display page of a client of a first application logged in using a first account is slid to a target display area, where the target display area is used to display one or more second applications pushed to the first account, and the one or more applications are part or all of a second application set, where the second application set is a set of second applications in which a first group of target accounts operates through the first application, the first group of target accounts is a group of accounts related to the first account, which is determined from an account set of the first application, and a first service resource operated by the first account in the first application is similar to a second service resource operated by the first group of target accounts in the first application;

(2) A display unit 1104, configured to push, in the target display area, one or more second applications to the first account.

Optionally, in this embodiment, the service resource pushing device may be located in a terminal, and may be, but not limited to, applied to a pushing process of a game application, where the service resource pushing method may be, but not limited to, applied to an online video media application, a news application, a social application, and other applications with service resource operation functions, so as to implement pushing, according to operation service resources, a second application interested by a user. For example, according to the similarity of video resources watched by the user in the online video media application, the game application clicked and downloaded by the user similar to the user through the online video media application is recommended to the user, which is just an example and is not limited in any way in the present embodiment.

Alternatively, in the present embodiment, the second determining unit 1102 may be used to perform the aforementioned step S602, and the display unit 1104 may be used to perform the aforementioned step S604.

As an optional technical solution, the apparatus further includes:

(1) The sending unit is used for sending a request message to the server after determining that a display page of a client of a first application logged in by using a first account is slid down to a target display area, wherein the request message is used for requesting to acquire application information of one or more second applications pushed to the first account;

(2) And the receiving unit is used for receiving a response message sent by the server and used for responding to the request message, wherein the response message carries application information of one or more second applications.

As an optional technical solution, the apparatus further includes:

(1) The detection unit is used for detecting target operations executed on target applications in one or more second applications after the target display area pushes the one or more second applications to the first account;

(2) The jump unit is used for responding to the target operation and jumping to the target application under the condition that the terminal where the client is located is provided with the target application; or, if the terminal where the client is located is not provided with the target application, responding to the target operation and jumping to a downloading interface of the target application.

According to an embodiment of the present invention, there is also provided a storage medium. Alternatively, in the present embodiment, the storage medium may be located in a network device in a network. The network device may be a server or a terminal. The storage medium is configured to store program code for performing the service resource pushing method.

Alternatively, in the present embodiment, the storage medium is configured to store program code for performing the steps of:

s1, determining a first group of target accounts related to a first account in an account set of a first application, wherein a first business resource operated by the first account in the first application is similar to a second business resource operated by the first group of target accounts in the first application;

S2, acquiring a second application set of the first group of target accounts through the first application operation;

and S3, pushing part or all of the second application set to the first account.

s1, determining that a display page of a client of a first application logged in by using a first account is slid to a target display area, wherein the target display area is used for displaying one or more second applications pushed to the first account, the one or more applications are part or all of a second application set, the second application set is a set of second applications, which are operated by a first group of target accounts through the first application, the first group of target accounts are a group of accounts related to the first account, which are determined from an account set of the first application, and first business resources operated by the first account in the first application are similar to second business resources operated by the first group of target accounts in the first application;

s2, pushing one or more second applications to the first account in the target display area.

Those of ordinary skill in the art will appreciate that all or part of the steps in the various methods of the above embodiments may be implemented by a program for instructing a terminal device to execute in association with hardware, the program may be stored in a computer readable storage medium, and the storage medium may include: flash disk, ROM (Read-Only Memory), RAM (Random Access Memory ), magnetic or optical disk, and the like.

According to an embodiment of the present invention, an electronic device for implementing the service resource pushing method is further provided, and optionally, in this embodiment, the electronic device may be located in a network device in a network. As shown in fig. 12, the electronic device includes: one or more (only one is shown) processors 1202, memory 1204, and transmission 1206. The memory has stored therein a computer program, the processor being arranged to perform the steps of any of the method embodiments described above by means of the computer program.

Alternatively, in the present embodiment, the above-described processor may be configured to execute the following steps by a computer program:

and S3, pushing part or all of the second application set to the first account.

It will be appreciated by those skilled in the art that the structure shown in fig. 12 is merely illustrative, and the electronic device may be a server (e.g., a cloud server), a smart phone (e.g., an Android phone, an iOS phone, etc.), a tablet computer, a palm computer, a MID (Mobile Internet Devices, a mobile internet device), a PAD, etc. Fig. 12 is not limited to the structure of the electronic device. For example, the electronic device may also include more or fewer components (e.g., network interfaces, display devices, etc.) than shown in FIG. 12, or have a different configuration than shown in FIG. 12.

The memory 1204 may be used to store software programs and modules, such as program instructions/modules corresponding to the service resource pushing method in the embodiment of the present invention, and the processor 1202 executes the software programs and modules stored in the memory 1204, thereby executing various functional applications and data processing, that is, implementing the service resource pushing method described above. Memory 1204 may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory 1204 may further include memory remotely located relative to the processor 1202, which may be connected to the processor 1202 via a network. Examples of such networks include, but are not limited to, the internet, intranets, local area networks, mobile communication networks, and combinations thereof.

The transmission device 1206 is configured to receive or transmit data via a network. Specific examples of the network described above may include wired networks and wireless networks. In one example, the transmission means 1206 comprises a NIC (Network Interface Controller, network adapter) that can be connected to other network devices and routers via network lines to communicate with the internet or a local area network. In one example, the transmission device 1206 is an RF (Radio Frequency) module that is configured to communicate wirelessly with the internet.

The foregoing embodiment numbers of the present application are merely for the purpose of description, and do not represent the advantages or disadvantages of the embodiments.

The integrated units in the above embodiments may be stored in the above-described computer-readable storage medium if implemented in the form of software functional units and sold or used as separate products. Based on such understanding, the technical solution of the present application may be embodied in essence or a part contributing to the prior art or all or part of the technical solution in the form of a software product stored in a storage medium, comprising several instructions for causing one or more computer devices (which may be personal computers, servers or network devices, etc.) to perform all or part of the steps of the method described in the embodiments of the present application.

In the foregoing embodiments of the present application, the descriptions of the embodiments are emphasized, and for a portion of this disclosure that is not described in detail in this embodiment, reference is made to the related descriptions of other embodiments.

In several embodiments provided by the present application, it should be understood that the disclosed client may be implemented in other manners. The above-described embodiments of the apparatus are merely exemplary, and the division of the units, such as the division of the units, is merely a logical function division, and may be implemented in another manner, for example, multiple units or components may be combined or may be integrated into another system, or some features may be omitted, or not performed. Alternatively, the coupling or direct coupling or communication connection shown or discussed with each other may be through some interfaces, units or modules, or may be in electrical or other forms.

The units described as separate units may or may not be physically separate, and units shown as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

In addition, each functional unit in the embodiments of the present invention may be integrated in one processing unit, or each unit may exist alone physically, or two or more units may be integrated in one unit. The integrated units may be implemented in hardware or in software functional units.

The foregoing is merely a preferred embodiment of the present invention and it should be noted that modifications and adaptations to those skilled in the art may be made without departing from the principles of the present invention, which are intended to be comprehended within the scope of the present invention.

Claims

1. The service resource pushing method is characterized by comprising the following steps:

determining a first group of target accounts related to a first account in a first application in an account set of the first application, wherein a first business resource operated by the first account in the first application is similar to a second business resource operated by the first group of target accounts in the first application;

Acquiring a second application set of the first group of target accounts operated by the first application;

determining a preference value of each target account in the first group of target accounts for the second application according to the operation times of each target account for the second application in the second application set through the first application, wherein the preference value is a result obtained by carrying out weighted summation on the operation times through target weights;

according to the vector similarity between the fifth vector used for representing each target account and the sixth vector used for representing the first account, weighting and summing the preference values of the second application by each target account to obtain a second target value corresponding to the second application;

sorting the second applications in the second application set according to the order of the second target values corresponding to the second applications in the second application set from big to small;

and selecting part or all of the second applications from the second application set according to the ordering result obtained by ordering, and pushing the second applications to the first account.

2. The method of claim 1, wherein determining the first set of target accounts related to the first account in the set of accounts in the first application comprises:

determining a first vector according to first operation information of a first target operation executed by the first account on the first service resource and a first label corresponding to the first service resource;

determining a second vector according to second operation information of the first target operation executed by the account in the account set on the second service resource and a second label corresponding to the second service resource;

and determining the first group of target accounts from the account set according to the vector similarity between the second vector and the first vector.

3. The method of claim 2, wherein determining the first vector based on the first operation information of the first target operation performed on the first service resource by the first account and the first label corresponding to the first service resource comprises:

determining a first tag set corresponding to the first account, wherein the first tag set is obtained by adding the first tag corresponding to the target service resource to the first tag set under the condition that the first account executes the first target operation on the target service resource of the first service resource;

Determining a first target value corresponding to a target label in the first label set, wherein the first target value is positively correlated with a quotient of the number of target labels contained in the first label set and the total number of labels contained in the first label set, is positively correlated with a quotient of the total number of accounts contained in the account set and the number of accounts contained in a second group of target accounts, and the target label is included in the second label corresponding to the account in the second group of target accounts;

and determining the first vector according to the first target value, wherein the element value of the target element corresponding to the target label in the first vector is the first target value.

4. The method of claim 2, wherein determining the first set of target accounts from the set of accounts based on vector similarity between the second vector and the first vector comprises:

determining the first group of target accounts in the account set, wherein the vector similarity between the second vector corresponding to each target account in the first group of target accounts and the first vector is larger than a preset threshold; or alternatively, the process may be performed,

And determining the first M target accounts in the account set as the first group of target accounts according to the sequence of the vector similarity between the second vector and the first vector corresponding to each target account in the account set from large to small, wherein M is a natural number.

5. The method of claim 2, wherein determining the first set of target accounts from the set of accounts based on the vector similarity between the second vector and the first vector comprises:

determining the first group of target accounts in the account set at least according to the vector similarity between a third vector and the first vector and the vector similarity between a fourth vector and the third vector, wherein the third vector is a vector corresponding to a second account with a first-order neighbor relation with the first account in a target relationship network, the fourth vector is a third account with a second-order neighbor relation with the first account through the second account in the target relationship network, the target relationship network is a non-directional weighted relationship network, the nodes of the target relationship network represent accounts in the account set, and the weight on the edge between two nodes in the target relationship network is the vector similarity between the second vectors corresponding to the two nodes.

6. The method according to any one of claim 1 to 5, wherein,

determining, in an account set of a first application, a first set of target accounts related to a first account in the account set includes: determining a first group of target accounts related to a first account in a first account set in an account set of a first application, wherein a first video resource operated by the first account in the first application is similar to a second video resource operated by the first group of target accounts in the first application;

the obtaining the second application set operated by the first group of target accounts through the first application comprises the following steps: and acquiring a game application set operated by the first group of target accounts through the first application.

7. The service resource pushing method is characterized by comprising the following steps:

determining that a display page of a client of a first application logged in by using a first account is slid to a target display area, wherein the target area is used for displaying one or more second applications pushed to the first account, the one or more applications are part or all of a second application set, the second application set is a set of second applications, a first group of target accounts is operated by the first application, the first group of target accounts is a group of accounts related to the first account, which is determined from an account set of the first application, and a first service resource operated by the first account in the first application is similar to a second service resource operated by the first group of target accounts in the first application;

The one or more second applications pushed to the first account are in the target display area;

wherein the one or more second applications pushed to the first account include:

8. The method of claim 7, wherein after determining that the display page of the client of the first application logged in using the first account is slid to the target display area, the method further comprises:

sending a request message to a server, wherein the request message is used for requesting to acquire application information of the one or more second applications pushed to the first account;

and receiving a response message sent by the server and used for responding to the request message, wherein the response message carries the application information of the one or more second applications.

9. The method of claim 7 or 8, wherein after the one or more second applications pushed by the target display area to the first account, the method further comprises:

under the condition that the terminal where the client is located is provided with the target application, responding to the target operation and jumping to the target application; or alternatively, the process may be performed,

10. A service resource pushing device, comprising:

a first determining unit, configured to determine, in an account set of a first application, a first group of target accounts related to a first account in the account set, where a first service resource operated by the first account in the first application is similar to a second service resource operated by the first group of target accounts in the first application;

the acquisition unit is used for acquiring a second application set of the first group of target accounts operated by the first application;

a pushing unit, configured to push part or all of the second application set to the first account;

the pushing unit comprises:

the ordering module is used for ordering the second application in the second application set according to third operation information of second target operations executed by each target account in the first group of target accounts through the first application on the second application in the second application set;

the selecting module is used for selecting part or all of the second applications from the second application set according to the ordering result obtained by ordering and pushing the second applications to the first account;

The ordering module comprises:

a seventh determining submodule, configured to determine a preference value of each target account to the second application according to the number of operations of each target account to the second target operation performed by the second application in the second application set through the first application, where the preference value is a result obtained by performing weighted summation on the number of operations through a target weight;

a summation sub-module, configured to weight and sum the preference values of the second application for each target account according to a vector similarity between a fifth vector for representing each target account and a sixth vector for representing the first account, so as to obtain a second target value corresponding to the second application;

and the sequencing sub-module is used for sequencing the second applications in the second application set according to the sequence from the big to the small of the second target value corresponding to the second applications in the second application set.

11. A service resource pushing device, comprising:

a second determining unit, configured to determine that a display page of a client of a first application logged in using a first account is slid to a target display area, where the target area is configured to display one or more second applications pushed to the first account, where the one or more applications are part or all of a second application set, where the second application set is a set of second applications that a first group of target accounts operates through the first application, and the first group of target accounts is a group of accounts related to the first account, which is determined from an account set of the first application, and a first service resource operated by the first account in the first application is similar to a second service resource operated by the first group of target accounts in the first application;

The display unit is used for pushing the one or more second applications to the first account in the target display area;

12. A storage medium having a computer program stored therein, wherein the computer program is arranged to perform the method of any of claims 1 to 6, or 7 to 9, when run.

13. An electronic device comprising a memory and a processor, characterized in that the memory has stored therein a computer program, the processor being arranged to execute the method according to any of the claims 1 to 6, or 7 to 9 by means of the computer program.