CN103138954B - Method, system and recommendation server for pushing recommended items - Google Patents

Abstract

The invention discloses a kind of method for pushing of recommendation items, system and recommendation server, when determining the operation trigger recommendation event of active user, according to active user to each targeted customer in the Generalization bounds of current item to be recommended and the buddy list of active user to the reception strategy of current item to be recommended, produce candidate target user list; To determine according to candidate target user list and after the list of targeted subscribers returned, in list of targeted subscribers, each targeted customer recommends current item to be recommended receiving active user; For sending the active user of recommendation items, the targeted customer that can reduce to adopting probability low pushes recommendation items, has saved recommendation resource, and improving recommendation items by the basis of success rate of adopting, adds the flexibility that active user pushes; For pushed targeted customer, avoid and receive a large amount of uninterested recommendation items, saved recommendation resource, improve the probability adopted to current item to be recommended.

Description

Method, system and recommendation server for pushing recommended items

technical field

The invention relates to the field of business support, in particular to a method, system and recommendation server for pushing recommended items.

Background technique

With the rise of e-commerce and social media sites, recommendation technology has been widely used in many popular applications such as Taobao, Douban, Google News, Amazon, and Dianping.com. Traditional recommendation systems are mainly divided into two categories: content-based recommendation systems and collaborative filtering systems. In a content-based recommendation system, the input data is processed as individual user profiles, and each user profile is generally represented as a feature vector, and the recommendation candidate information is similarly processed, and then the similarity with the target user's profile is performed Calculated, the candidate information closest to the user profile is pushed to the user as a recommended item. In a collaborative filtering system, user behavior data is used to calculate the similarity between users or recommendation candidate information, and the recommendation results are further weighted according to the similarity.

In a typical social network-based recommendation, users can send their favorite recommended items to other users in the social network and receive recommended items pushed by other users from other users. However, due to the complexity of social network , a user’s friends may not be interested in the recommended items he pushes, and at the same time, a user will also receive a large number of recommended items that he is not interested in, resulting in a waste of recommended resources and reducing the success of recommended items being adopted rate; and when the user repeatedly receives recommended items from a friend, the possibility of adopting the recommended item will gradually decrease, reducing the success rate of the overall recommended item being adopted.

Contents of the invention

Embodiments of the present invention provide a method, system, and recommendation server for pushing recommended items, which are used to solve the problems of waste of recommended resources caused by existing pushing methods and low success rate of adoption of recommended items.

A method for pushing recommended items provided by an embodiment of the present invention includes:

When it is determined that the operation of the current user triggers a recommendation event, acquiring the friend list of the current user according to the identifier of the current user;

Generate a list of candidate target users according to the current user's recommendation strategy for the current item to be recommended and the receiving strategy of each target user in the friend list for the current item to be recommended;

presenting the candidate target user list to the current user, receiving the target user list determined and returned by the current user according to the candidate target user list;

Pushing the current to-be-recommended item to each target user in the target user list according to the instruction sent by the current user for confirming sending of the recommendation.

A recommendation server provided by an embodiment of the present invention includes:

An acquisition module, configured to acquire a friend list of the current user according to the identifier of the current user when determining that the operation of the current user triggers a recommendation event;

A calculation module, configured to generate a list of candidate target users according to the current user's recommendation strategy for the current item to be recommended and the receiving strategy of each target user in the friend list for the current item to be recommended;

A confirmation module, configured to present the candidate target user list to the current user, and receive the target user list determined and returned by the current user according to the candidate target user list;

The recommending module is configured to push the current to-be-recommended item to each target user in the target user list according to the instruction sent by the current user for confirming the recommendation.

A system for pushing recommended items provided by an embodiment of the present invention includes: a recommendation server, a business server, and an information release platform server;

The recommendation server is configured to obtain a friend list of the current user according to the identification of the current user when determining that the operation of the current user triggers a recommendation event; according to the current user's recommendation strategy for the current item to be recommended and the The receiving strategy of each target user in the friend list for the current item to be recommended generates a candidate target user list; presents the candidate target user list to the current user, receives the current user's determination according to the candidate target user list and returns The target user list; according to the instruction sent by the current user to confirm sending the recommendation, push the current item to be recommended to each target user in the target user list;

The service server is used to provide the user's friend information and recommended item information;

The information release platform server is configured to release the currently to-be-recommended item pushed by the recommendation server to the receiving platform of the target user.

The beneficial effects of the embodiments of the present invention include:

The embodiment of the present invention provides a method, system, and recommendation server for pushing recommended items. When it is determined that the current user's operation triggers a recommendation event, according to the current user's recommendation strategy for the current item to be recommended and each target in the current user's friend list The user's receiving strategy for the current item to be recommended generates a candidate target user list; after receiving the target user list determined and returned by the current user according to the candidate target user list, recommend the current to-be-recommended item to each target user in the target user list. The method for pushing recommended items provided by the embodiment of the present invention comprehensively refers to the recommendation strategy and receiving strategy between the current user and the target user when generating the target user list of the current to-be-recommended item, and screens out target users suitable for the current to-be-recommended item , and finally through the current user to confirm each target user in the target user list, for the current user who sends the recommended item, it is possible to reduce the push of recommended items to the target users with low probability of adoption, saving recommendation resources, and improving the recommended items. On the basis of the success rate of adoption, the flexibility of the current user's push is increased; for the target users to be pushed, it avoids receiving a large number of uninterested recommendation items, saves recommendation resources, and improves the adoption of current items to be recommended probability.

Description of drawings

FIG. 1 is a flowchart of a method for pushing recommended items provided by an embodiment of the present invention;

FIG. 2 is a flow chart of obtaining a recommendation strategy provided by an embodiment of the present invention;

FIG. 3 is a flow chart of acquiring and receiving policies provided by an embodiment of the present invention;

FIG. 4 is a schematic structural diagram of a recommendation server provided by an embodiment of the present invention;

FIG. 5 is an architecture diagram of a system for pushing recommended items of an example provided by an embodiment of the present invention;

FIG. 6 is a schematic structural diagram of a recommendation server in an example provided by an embodiment of the present invention.

Detailed ways

The specific implementation manners of the method, system and recommendation server for pushing recommended items provided by the embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

A method for pushing recommended items provided by an embodiment of the present invention, as shown in FIG. 1 , the specific process includes the following steps:

S101. When it is determined that the operation of the current user triggers a recommendation event, obtain the friend list of the current user according to the identifier of the current user;

S102. Generate a list of candidate target users according to the current user's recommendation strategy for the current item to be recommended and each target user's receiving strategy for the current item to be recommended in the buddy list;

S103. Present the candidate target user list to the current user, and receive the target user list determined and returned by the current user according to the candidate target user list;

S104. According to the instruction sent by the current user for confirming and sending the recommendation, push the currently to-be-recommended item to each target user in the target user list.

The specific implementation of the above steps will be described in detail below.

In the above step S101, the operation of the current user triggering the recommendation event may include but not limited to the following events: (1) the current user completes the payment of the online order; (2) the current user completes the download of the setting software; (3) The current user finishes commenting on the set information; (4) The current user chooses to share the information with his friends. When the current user completes the above events, a recommendation event will be triggered, and the system will obtain the current user’s friend list according to the current user’s identification. Import the friend list of the current user, and the source of the friend list is not limited here.

Preferably, in the above step S102, the specific process of generating the list of candidate target users may include the following steps:

First, according to the current user's recommendation strategy for the current item to be recommended, each target user in the friend list is screened to obtain a list of candidate target users that meet the recommendation strategy;

Then, according to the receiving strategy of each target user in the friend list for the current item to be recommended, the target users in the list of candidate target users conforming to the recommendation strategy are screened to obtain a list of candidate target users conforming to the receiving strategy.

The candidate target user list obtained through the above two screenings not only conforms to the current user's recommendation strategy, but also conforms to the target user's receiving strategy. In this way, for the current user who sends the recommended item, it is possible to prevent the current user from pushing the recommended item to For the target users who are not interested in this recommendation item, the recommendation resources are saved. For the target users who are pushed, it avoids receiving a large number of uninterested recommendation items, saves recommendation resources, and improves the adoption of current items to be recommended. probability.

Specifically, the acquisition of the user's recommendation strategy and receiving strategy in the above method provided by the embodiment of the present invention can be achieved through the following steps:

For the recommendation strategy for acquiring users, as shown in Figure 2, the following steps may be included:

S201. For each online user, obtain a recommendation strategy set by the user for each currently to-be-recommended item;

S202. According to the user's recommendation strategy and the receiving strategy of each target user in the user's friend list for the current item to be recommended, predict the proportion of the current item to be recommended by the target user in the friend list; for example: predict how many friends the user's recommendation can be recommended by Received, and the likelihood of being adopted by those friends;

S203. Presenting the prediction result to the user;

S204. When receiving the recommendation strategy confirmation instruction sent by the user, save the recommendation strategy set by the user;

S205. When receiving a recommendation strategy update instruction sent by the user, save the recommendation strategy updated by the user.

Similarly, the receiving strategy for acquiring users is similar to the above-mentioned recommendation strategy process for acquiring users, as shown in Figure 3, and can be implemented through the following steps:

S301. For each online user, obtain a receiving policy set by the user for each currently to-be-recommended item;

S302. Predict the number of recommendations that the user will receive within a certain period of time according to the user's receiving strategy and the recommendation strategy of each target user in the user's friend list for the current item to be recommended;

S303. Presenting the prediction result to the user;

S304. When receiving the receiving policy confirmation instruction sent by the user, save the receiving policy set by the user;

S305. When receiving the receiving policy update instruction sent by the user, save the user's updated receiving policy.

During specific implementation, the steps S201-S205 of acquiring the user's recommendation strategy and the steps S301-S305 of acquiring the user's receiving strategy can be performed simultaneously or separately, which are not limited here.

Preferably, in step S103 of the above method provided by the embodiment of the present invention, before receiving the target user list determined and returned by the current user according to the candidate target user list, when the current user modifies its recommendation strategy, the candidate target user list will be updated. target user list, and present the updated candidate target user list to the current user.

Specifically, in the above method provided by the embodiment of the present invention, the recommendation strategy or reception strategy used may be one of the following strategies:

Black and white list strategy, circle of influence strategy, preference strategy, and a combined strategy of circle of influence strategy and preference strategy; among them, the circle of influence strategy is based on the number of times the current user recommends each target user in his friend list and each target user's current Policy set by the user's recommendation adoption.

Each strategy is described in detail below.

Black and white list strategy: Divide the friend list into two lists, all friends in the white list will be the target users, and all friends in the black list will be removed.

Circle of Influence Strategy: It consists of three basic sub-strategies, which can be connected by logical "AND" or logical "OR" to achieve different filtering effects. Specifically, the sub-strategies are: 1) The current user sends to the Friends who have adopted their recommended items within the time interval send new recommended items; 2) The current user sends new recommended items to friends whose accumulated recommended times within the time interval are less than k; 3) The current user sends new recommended items to friends who have recently received n Friends who have recommended items and adopted at least one of them send new recommended items.

For example: Assume that only 1) or 2) strategies are adopted, the time interval is set to 1 week, and the upper limit k of the cumulative number of recommendations sent is set to 7, then, if the number of recommendations sent by user A to friend B within a week has not reached 7 , or more than 7 but friend B has adopted one of them, then friend B will be the target user, otherwise friend B will be removed from the target users.

Utilizing the circle of influence strategy can prevent the current user from sending a large number of items to be recommended to those friends who continue to pay no attention to their recommendations. At the same time, it is also possible to test the degree of attention of friends to their recommendations by sending a small amount.

Preference degree strategy: By setting a threshold, friends whose preference for the current item to be recommended is greater than the threshold are used as target users. Since the calculation method of preference degree belongs to the prior art, it will not be described in detail here.

In a specific embodiment, the setting of the threshold can normalize the preference of friends according to certain principles (for example, linearly normalize to [0, 1] interval), and take the average preference of friends for the current items to be recommended as Threshold, or set a specific value as required, for example, using 0 as the threshold can filter out all friends who show preference for the current item to be recommended.

Preferably, step S104 in the method provided by the embodiment of the present invention pushes the current item to be recommended to each target user in the target user list according to the instruction sent by the current user to confirm sending the recommendation. A unique identifier is added to the access link of the item to be recommended currently. Through this identifier, it can be identified which user sent the recommended item clicked by the user. In this way, by recording the user's click on the recommended result and the conversion of the purchase, it will be used for future reference. Subsequent user setting policies are provided for reference.

Based on the same inventive concept, the embodiment of the present invention also provides a recommendation server and a system for pushing recommended items. Since the problem-solving principle of the server and the system is similar to that of the aforementioned method for pushing recommended items, the implementation of the server and the system Reference can be made to the implementation of the method, and repeated descriptions will not be repeated.

A recommendation server provided by an embodiment of the present invention, as shown in FIG. 4 , includes:

An acquisition module 401, configured to acquire the current user's friend list according to the current user's identifier when determining that the current user's operation triggers a recommendation event;

A calculation module 402, configured to generate a list of candidate target users according to the current user's recommendation strategy for the current item to be recommended and the receiving strategy of each target user in the buddy list for the current item to be recommended;

Confirmation module 403, configured to present the candidate target user list to the current user, and receive the target user list determined and returned by the current user according to the candidate target user list;

The recommendation module 404 is configured to push the currently to-be-recommended item to each target user in the target user list according to the instruction sent by the current user for confirming the recommendation.

Further, the above-mentioned server provided by the embodiment of the present invention, as shown in FIG. 4 , may also include: a policy saving module 405, for each online user, to acquire the recommendation policy set by the user for each currently to-be-recommended item and receive Strategy: According to the user's recommendation strategy and the receiving strategy of each target user in the user's friend list for the current item to be recommended, predict the proportion of target users in the friend list for the current item to be recommended; or according to the user's receiving strategy and the user's friends The recommendation strategy of each target user in the list predicts the number of recommendations that the user will receive within a certain period of time; presents the prediction result to the user; when receiving the recommendation strategy sent by the user or receiving a policy confirmation instruction, save the recommendation strategy set by the user or Receiving strategy, when receiving a recommendation strategy sent by a user or receiving a strategy update instruction, saving the user's updated recommendation strategy or receiving strategy.

Further, the calculation module 402 in the above server provided by the embodiment of the present invention is specifically used to screen each target user in the friend list according to the current user's recommendation strategy for the current item to be recommended, and obtain candidate targets that meet the recommendation strategy User list: According to the receiving strategy of each target user in the friend list for the current item to be recommended, the target users in the list of candidate target users conforming to the recommendation strategy are screened to obtain a list of candidate target users conforming to the receiving strategy.

Further, the confirmation module 403 in the server provided by the embodiment of the present invention is also used to update the candidate target user list when it is acquired that the current user modifies the recommendation policy before receiving the target user list confirmed and returned by the current user according to the candidate target user list. target user list, and present the updated candidate target user list to the current user.

The embodiment of the present invention also provides a system for pushing recommended items, including: a recommendation server, a business server, and an information release platform server;

The recommendation server is used to obtain the friend list of the current user according to the identification of the current user when it is determined that the operation of the current user triggers a recommendation event; according to the current user's recommendation strategy for the current item to be recommended and the target users in the friend list for the current item to be recommended Item receiving strategy, generate a list of candidate target users; present the list of candidate target users to the current user, receive the list of target users determined and returned by the current user according to the list of candidate target users; send recommended instructions according to the confirmation sent by the current user, to the target Each target user in the user list pushes the current item to be recommended;

The business server is used to provide the user's friend information and recommendation item information;

The information publishing platform server is used to publish the currently to-be-recommended items pushed by the recommendation server to the receiving platform of the target user.

Further, the above-mentioned system provided by the embodiment of the present invention may also include: an external data source;

The recommendation server is also used to obtain the user's friend information from an external data source.

The above-mentioned system provided by the embodiment of the present invention is described below through specific examples, as shown in Figure 5, including a recommendation server, a business server, an information publishing platform server and an external data source (the information publishing platform server and the external data source are not shown in the figure) .

Wherein, the specific structure of the recommendation server is shown in Figure 6, which may include the following modules: a recommendation trigger control module 601, a recommendation target calculation module 602, a social network management module 603, a preference calculation module 604, a recommendation strategy control module 605, and a recommendation sending module. Module 606. Among them, the recommendation trigger control module 601, the social management module 603, the recommendation policy control module 605, and the recommendation sending module 606 may interact with the business server (such as an e-commerce platform) and external data sources (such as friends from social networking sites to obtain Web Service), information release platform server (such as Weibo release platform) information transfer occurs.

Specifically, the recommendation trigger control module 601 is configured to receive a recommendation trigger event from the service server, and submit a push request to the recommendation target calculation module 602 . Recommended triggering events are determined by the business needs of the system. Typical events include but are not limited to: 1) the user completes the payment of an online order; 2) the user downloads a certain software in the application store; 3) the user makes a comment on a piece of information ; 4) The user actively chooses to share the information with his friends.

The recommended target calculation module 602 is configured to receive a push request from the recommendation trigger control module 601. The push request includes a current user id and the id of the current item to be recommended. The recommended target calculation module 602 can manage according to these two identifications and social network The module 603, the preference degree calculation module 604, and the recommendation policy control module 605 interact to determine a target user group suitable for the current user and the current item to be recommended.

Specifically, the recommended target calculation module 602 includes: a social network request module 6021 , a policy request module 6022 , a preference request module 6023 , a historical behavior request module 6024 and a policy filter module 6025 .

In a specific application, the recommended target calculation module 602 calls the social network request module 6021 and the policy request module 6022 in sequence, and decides whether to call the preference request module 6023 and the history behavior request module 6024 according to the obtained policy to obtain the required user friends list, user historical behavior, and friend’s preference for the current item to be recommended. Afterwards, the recommendation target calculation module 602 invokes the policy filtering module 6025 to perform policy filtering on the user’s friend list to generate the final candidate target user list, which is provided to the recommendation sending Module 606 performs user confirmation and completes the sending.

The social network management module 603 is used to maintain the user's friend list, and when receiving the request from the recommendation target calculation module 602, returns the friend list according to the current user id.

Specifically, the social network management module 603 includes: an internal friend management module 6031 and an external friend management module 6032, which are respectively used to manage the user's friends inside the system and friends outside the system, wherein the internal friends refer to the user's use of the internal friends provided by the system. The function of making friends directly adds other users as friends; external friends refer to the friendships imported by users from data sources such as email address books, social networking sites, and chat tools. In a specific application, the external friend management module 6032 may also include an invitation confirmation function for inviting an external user to become an internal user, or it may also utilize APIs provided by open platforms such as Google OpenSocial and Facebook Friend Connect to directly communicate with external data sources. interact.

The preference calculation module 604 is used to calculate the preference degree of the target user for the current recommended item. This module receives the user's friend list from the recommended target calculation module 602 and the id of the current item to be recommended, and then returns all target user pairs in the friend list. The preference of the current recommended item, the preference data comes from the user behavior recorded by the business system (such as purchase, click, download, on-demand, etc.).

Specifically, the preference calculation module 604 includes: a content-based preference calculation module 6041 , a collaborative filtering-based preference calculation module 6042 , and a mixed strategy-based preference calculation module 6043 . The content-based preference calculation module 6041 calculates the user's preference for recommended items according to the text features of the user's historical access objects and the text features of the items to be recommended. Generally speaking, the algorithms adopted by this module need to represent the user's preferences and items to be recommended as attribute vectors. The collaborative filtering-based preference calculation module 6042 calculates the user's preference for the current item to be recommended based on the similarity of the user's historical access records and the access records of other users similar to the user. The hybrid policy-based preference calculation module 6043 combines the features of the content-based preference calculation module 6041 and the collaborative filtering-based preference calculation module 6042 to improve the comprehensiveness of preference calculation. During specific implementation, the preference degree of the target user in the friend list to the current item to be recommended can be calculated according to the preference degree request (including the identification of the current item to be recommended and the friend list) sent by the preference degree request module 6023, and the preference degree of the target user in the friend list can be calculated along with the friend list Return to the recommendation target calculation module 602 for subsequent policy filtering calculation.

The recommendation policy control module 605 is used for users to set corresponding recommendation policies and receiving policies according to their needs. These strategies include but are not limited to influence circle strategy, black and white list strategy, preference strategy, etc.

Specifically, the recommendation policy control module 605 includes: a black and white list policy module 6051, an influence circle policy module 6052, and a preference policy module 6053. Users can set three types of recommendation policies through the above three modules: black and white list policies, and influence circle policies , Preference strategy.

The recommendation sending module 606 is configured to send the recommended item to the corresponding target user group according to the output of the recommendation target calculation module 602 . This module includes a process for the recommendation to be edited and confirmed by the current user before being sent.

Specifically, the recommendation sending module 606 includes: an internal sending module 6061 and an external sending module 6062 . The internal sending module 6061 is used to display the recommended items on the display page in the system. When the recommended target user logs in to the system, he will see the content of the recommended items on the corresponding page. The external sending module 6062 is used to use the interface provided by the external system to send the recommended item to the display stand of the external system, wherein the typical external system means include email, microblog, and web service interface provided by social networking sites. Further, in order to track the user’s acceptance of the recommended item, the recommendation sending module 606 will add a unique identifier to the generated access link, through which it can be identified that the user x clicked on the recommended item of user y instead of user z of. The business system specifically records the user's click/purchase conversion on recommended items, and provides these data to the system through the interface of the recommendation strategy control module 605 . Finally, out of respect for user privacy, the recommendation sending module in any embodiment should provide a user confirmation function and allow the user to manually adjust the final target user list.

The above-mentioned system for pushing recommended items provided by the embodiments of the present invention is only an example, and the system for pushing recommended items may also include other forms during specific implementation, which will not be described in detail here.

Through the above description of the implementation manners, those skilled in the art can clearly understand that the embodiments of the present invention can be implemented by hardware, or by means of software plus a necessary general hardware platform. Based on such understanding, the technical solutions of the embodiments of the present invention can be embodied in the form of software products, which can be stored in a non-volatile storage medium (which can be CD-ROM, U disk, mobile hard disk, etc.), Several instructions are included to make a computer device (which may be a personal computer, a server, or a network device, etc.) execute the methods described in various embodiments of the present invention.

Those skilled in the art can understand that the drawing is only a schematic diagram of a preferred embodiment, and the modules or processes in the drawing are not necessarily necessary for implementing the present invention.

Those skilled in the art can understand that the modules in the device in the embodiment can be distributed in the device in the embodiment according to the description in the embodiment, or can be located in one or more devices different from the embodiment according to corresponding changes. The modules in the above embodiments can be combined into one module, and can also be further split into multiple sub-modules.

The serial numbers of the above embodiments of the present invention are for description only, and do not represent the advantages and disadvantages of the embodiments.

The embodiment of the present invention provides a method, system, and recommendation server for pushing recommended items. When it is determined that the current user's operation triggers a recommendation event, according to the current user's recommendation strategy for the current item to be recommended and each target in the current user's friend list The user's receiving strategy for the current item to be recommended generates a candidate target user list; after receiving the target user list determined and returned by the current user according to the candidate target user list, recommend the current to-be-recommended item to each target user in the target user list. The method for pushing recommended items provided by the embodiment of the present invention comprehensively refers to the recommendation strategy and receiving strategy between the current user and the target user when generating the target user list of the current to-be-recommended item, and screens out target users suitable for the current to-be-recommended item , and finally through the current user to confirm each target user in the target user list, for the current user who sends the recommended item, it is possible to reduce the push of recommended items to the target users with low probability of adoption, saving recommendation resources, and improving the recommended items. On the basis of the success rate of adoption, the flexibility of the current user's push is increased; for the target users to be pushed, it avoids receiving a large number of uninterested recommendation items, saves recommendation resources, and improves the adoption of current items to be recommended probability.

Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and equivalent technologies thereof, the present invention also intends to include these modifications and variations.

Claims (9)

1. A method for pushing recommended items, comprising: When it is determined that the operation of the current user triggers a recommendation event, acquiring the friend list of the current user according to the identifier of the current user; Generate a list of candidate target users according to the current user's recommendation strategy for the current item to be recommended and the receiving strategy of each target user in the friend list for the current item to be recommended; presenting the candidate target user list to the current user, receiving the target user list determined and returned by the current user according to the candidate target user list; Push the current to-be-recommended item to each target user in the target user list according to the instruction sent by the current user to confirm sending the recommendation; wherein, before determining that the operation of the current user triggers a recommendation event, it also includes: for each online According to the user's recommendation strategy and receiving strategy for each item to be recommended currently set by the user; according to the user's recommendation strategy and the receiving strategy of each target user in the user's friend list for the current item to be recommended, predict the friend The ratio of target users in the list to the current items to be recommended; or predict the number of recommendations that the user will receive within a certain period of time according to the user’s receiving strategy and the recommendation strategies of each target user in the user’s friend list ;Present the prediction result to the user; when receiving the recommendation strategy or receiving strategy confirmation instruction sent by the user, save the recommendation strategy or receiving strategy set by the user, and when receiving the recommendation strategy or receiving strategy sent by the user When the instruction is updated, the user's updated recommendation strategy or receiving strategy is saved. 2. The method according to claim 1, wherein, according to the current user's recommendation strategy for the current item to be recommended and the receiving strategy of each target user in the friend list for the current item to be recommended, candidate target users are generated list, including: According to the current user's recommendation strategy for the current item to be recommended, filter each target user in the friend list to obtain a list of candidate target users conforming to the recommendation strategy; According to the receiving strategy of each target user in the friend list for the current item to be recommended, the target users in the list of candidate target users conforming to the recommendation strategy are screened to obtain a list of candidate target users conforming to the receiving strategy. 3. The method according to claim 2, wherein the recommendation strategy or the receiving strategy is one of the following strategies: black and white list strategy, circle of influence strategy, preference strategy, and a combined strategy of circle of influence strategy and preference strategy; A strategy set by the target user for the adoption of the current user's recommendation. 4. The method according to any one of claims 1-3, wherein before receiving the target user list determined and returned by the current user according to the candidate target user list, further comprising: When it is acquired that the current user modifies the recommendation strategy, the list of candidate target users is updated, and the updated list of candidate target users is presented to the current user. 5. A recommendation server, characterized in that it comprises: An acquisition module, configured to acquire a friend list of the current user according to the identifier of the current user when determining that the operation of the current user triggers a recommendation event; A calculation module, configured to generate a list of candidate target users according to the current user's recommendation strategy for the current item to be recommended and the receiving strategy of each target user in the friend list for the current item to be recommended; A confirmation module, configured to present the candidate target user list to the current user, and receive the target user list determined and returned by the current user according to the candidate target user list; A recommendation module, configured to push the current item to be recommended to each target user in the target user list according to the instruction sent by the current user to confirm and send the recommendation; The strategy preservation module is used for each online user to obtain the recommendation strategy and receiving strategy set by the user for each currently to-be-recommended item; according to the recommendation strategy of the user and the friend list of the user, the The receiving strategy of the recommended item, predicting the proportion of the target user in the friend list to the current item to be recommended; or predicting the user's receiving strategy and the recommendation strategy of each target user in the user's friend list according to the receiving strategy of the user The number of recommendations that will be received within a certain period of time; present the prediction results to the user; when receiving the recommendation strategy or receiving strategy confirmation instruction sent by the user, save the recommendation strategy or receiving strategy set by the user, and when receiving When the user sends a recommendation strategy or receiving strategy update instruction, save the user's updated recommendation strategy or receiving strategy. 6. The server according to claim 5, wherein the calculation module is specifically configured to filter each target user in the friend list according to the current user's recommendation strategy for the current item to be recommended, Obtaining a list of candidate target users conforming to the recommendation strategy; screening the target users in the list of candidate target users conforming to the recommendation strategy according to the receiving strategy of each target user in the friend list for the current item to be recommended, A list of candidate target users conforming to the receiving policy is obtained. 7. The server according to claim 5 or 6, wherein the confirmation module is further configured to, before receiving the target user list confirmed and returned by the current user according to the candidate target user list, when the When the current user modifies the recommendation strategy, the list of candidate target users is updated, and the updated list of candidate target users is presented to the current user. 8. A push system for recommended items, comprising: a recommendation server, a business server and an information release platform server; The recommendation server is configured to obtain a friend list of the current user according to the identification of the current user when determining that the operation of the current user triggers a recommendation event; according to the current user's recommendation strategy for the current item to be recommended and the The receiving strategy of each target user in the friend list for the current item to be recommended generates a candidate target user list; presents the candidate target user list to the current user, receives the current user's determination according to the candidate target user list and returns The target user list; according to the instruction sent by the current user to confirm sending the recommendation, push the current item to be recommended to each target user in the target user list; Wherein, before the recommendation server determines that the operation of the current user triggers the recommendation event, it is also used to: for each online user, obtain the recommendation strategy and receiving strategy set by the user for each item currently to be recommended; Strategy and the receiving strategy of each target user in the user's friend list for the current item to be recommended, predicting the proportion of the target user in the friend list for the current item to be recommended; or according to the user's receiving strategy and the user's The recommendation strategy of each target user in the friend list, predict the number of recommendations that the user will receive within a certain period of time; present the prediction results to the user; when receiving the recommendation strategy sent by the user or receiving a policy confirmation instruction , saving the recommendation strategy or reception strategy set by the user, and saving the recommendation strategy or reception strategy updated by the user when receiving the recommendation strategy or reception strategy update instruction sent by the user; The service server is used to provide the user's friend information and recommended item information; The information release platform server is configured to release the currently to-be-recommended item pushed by the recommendation server to the receiving platform of the target user. 9. The system of claim 8, further comprising: an external data source; The recommendation server is further configured to obtain the user's friend information from the external data source.