CN112905879B

CN112905879B - Recommendation method, recommendation device, server and storage medium

Info

Publication number: CN112905879B
Application number: CN202110127366.5A
Authority: CN
Inventors: 程波波
Original assignee: Beijing Dajia Internet Information Technology Co Ltd
Current assignee: Beijing Dajia Internet Information Technology Co Ltd
Priority date: 2021-01-29
Filing date: 2021-01-29
Publication date: 2024-03-19
Anticipated expiration: 2041-01-29
Also published as: CN112905879A

Abstract

The disclosure relates to a recommendation method, a recommendation device, a server and a storage medium, and belongs to the technical field of Internet. According to the technical scheme provided by the embodiment of the disclosure, the recommending efficiency of the object can be determined according to the number of the resources of the object with the target parameter larger than the target threshold and the number of the resources of any object in the target object set, and as the recommending efficiency is positively related to the number of the resources obtained when recommending the object, the greater the recommending efficiency of the object is, the greater the number of the resources obtained when recommending the object is, and further whether the target parameter of the object is improved is judged by judging whether the recommending efficiency reaches the first threshold or not, the object with higher recommending efficiency can be determined, more resources can be obtained, and the recommending efficiency and the maximization of the obtained number of the resources are realized.

Description

Recommendation method, recommendation device, server and storage medium

Technical Field

The disclosure relates to the technical field of internet, and in particular relates to a recommendation method, a recommendation device, a recommendation server and a recommendation storage medium.

Background

With the rapid development of internet technology and the gradual expansion of the scale of network users, services such as internet advertising, internet propaganda and the like have penetrated into aspects of life of people, and in the internet, service objects are usually required to be recommended for users. When recommending objects, it is generally necessary to sort parameters of a plurality of objects belonging to a certain index according to the index, and select the object with the highest sorting for recommendation.

To recommend objects preferentially, the parameters of the objects are typically increased, thereby increasing the chances that the objects are recommended. The number of resources is expended when recommending these objects preferentially, and therefore the total number of resources is typically set to constrain the number of resources. However, if the number of objects to be preferentially recommended is large, how to select a target object to be parameter-improved from among the large number of objects based on the total number of resources becomes a problem to be solved.

Disclosure of Invention

The present disclosure provides a recommendation method, apparatus, server and storage medium, which can determine an object with higher recommendation efficiency, and further can obtain a larger number of resources, thereby maximizing the recommendation efficiency and the obtained number of resources. The technical scheme of the present disclosure is as follows:

according to a first aspect of embodiments of the present disclosure, there is provided a recommendation method, the method including:

acquiring a first resource quantity of a first object and a second resource quantity of a second object, wherein the first object is determined based on an object with a target parameter larger than a target threshold value in a plurality of objects to be recommended, the second object is any object in a target object set, the first resource quantity is the resource quantity obtained when the first object is recommended, and the second resource quantity is the resource quantity obtained when the second object is recommended;

Determining a recommendation efficiency of the second object based on a resource gap between the first resource amount and the second resource amount, the recommendation efficiency being positively correlated with the second resource amount, the recommendation efficiency being negatively correlated with the resource gap;

and in response to the recommended efficiency of the second object reaching a first threshold, increasing the target parameter of the second object.

In the embodiment of the disclosure, according to the number of resources of an object whose target parameter is greater than the target threshold and the number of resources of any object in the target object set, the recommendation efficiency of the object can be determined, and since the recommendation efficiency is positively related to the number of resources obtained when recommending the object, the greater the recommendation efficiency of the object is, the greater the number of resources obtained when recommending the object is, and further, whether to improve the target parameter of the object is determined by determining whether the recommendation efficiency reaches the first threshold, the object with higher recommendation efficiency can be determined, and further, the number of resources with higher recommendation efficiency can be obtained, thereby realizing maximization of the recommendation efficiency and the obtained number of resources.

In some embodiments, after the target parameter of the second object is increased in response to the recommended efficiency of the second object reaching the first threshold, the method further comprises:

Selecting an object with a target parameter larger than the target threshold value from the second object and the plurality of objects to be recommended based on the target parameter after the second object is improved;

recommending the object with the target parameter being larger than the target threshold, and recording the amount of resources consumed when recommending the object with the target parameter being larger than the target threshold.

According to the target parameters of the second object after being improved and the target parameters of the plurality of objects to be recommended, selecting the objects with the target parameters larger than the target threshold value for recommendation, and ensuring the recommendation effect of the objects while realizing the priority recommendation of some objects, so that the number of resources obtained during the recommendation of the objects is improved.

In some embodiments, the increasing the target parameter of the second object in response to the recommended efficiency of the second object reaching a first threshold comprises:

and in response to the recommendation efficiency of the second object reaching a first threshold, and the amount of resources consumed by the currently recommended object meeting a target condition, increasing the target parameter of the second object.

When judging whether to improve the target parameters of the second object, not only the recommendation efficiency and the first threshold value but also the amount of resources consumed by the recommended object are considered, so as to avoid excessive amount of resources consumed.

In some embodiments, before the obtaining the first amount of resources of the first object and the second amount of resources of the second object, the method further comprises:

uniformly dividing the total number of resources in a target time period according to a plurality of sub-time periods in the target time period to obtain the number of resources corresponding to each sub-time period, so that the total number of the resources in the plurality of sub-time periods is equal to the total number of the resources, wherein the total number of the resources represents the total number of the resources consumed by recommending objects in the target time period;

and for any one of the plurality of sub-time periods, adjusting the first threshold based on a difference value between the number of resources corresponding to the sub-time period and the number of resources consumed by the sub-time period to obtain the adjusted first threshold, wherein the first threshold is inversely related to the difference value.

The method not only can smooth the total number of resources in the target time period to each sub-time period of the target time period, but also can effectively control the speed of the process of improving the target parameters in the target time period by avoiding that a large number of objects are improved by improving the target parameters in the initial stage of the target time period, further can adjust the first threshold based on the real-time consumption of resources, can reduce the first threshold when the consumed resources are small in number and large in difference, can improve the target parameters of the second object as much as possible, can improve the first threshold when the consumed resources are large in number and small in difference, can reduce the opportunity that the second object is improved by improving the target parameters, and can avoid excessive consumed resources.

In some embodiments, the method further comprises:

and acquiring the consumed resource quantity of the plurality of objects belonging to the target publisher from the plurality of recommended objects in the target time period, if the consumed resource quantity is larger than a second threshold value, increasing the first threshold value of the plurality of objects belonging to the target publisher, and if the consumed resource quantity is smaller than or equal to the second threshold value, decreasing the first threshold value of the plurality of objects belonging to the target publisher.

For the publishers with larger total consumption of the resources, the first threshold value can be properly increased, and for the publishers with smaller total consumption of the resources, the first threshold value can be properly reduced, so that the fairness and the healthfulness of the priority recommendation are improved.

In some embodiments, the method further comprises:

selecting a third resource quantity of at least one third object belonging to the target field from the target object set, wherein the third resource quantity is the resource quantity obtained when recommending the third object;

determining a recommendation efficiency for the target area based on the first number of resources and the third number of resources of the at least one third object;

and in response to the recommended efficiency of the target area reaching the first threshold, increasing a target parameter of at least one third object included in the target area.

By selecting the number of resources of a plurality of objects belonging to the same field, and further determining the recommendation efficiency of the field, whether to improve the target parameters of the plurality of objects can be judged more quickly, the recommendation efficiency is not required to be determined for the objects one by one, and the execution efficiency of the whole recommendation process is improved.

In some embodiments, the determining the recommendation efficiency for the target area based on the first number of resources and the third number of resources for the at least one third object comprises:

determining a target resource sum for the target domain based on a third number of resources for the at least one third object;

determining a sum of resource gaps for the target area based on the resource gaps between the first number of resources and the third number of resources of the at least one third object;

and determining the recommendation efficiency of the target field based on the target resource sum and the resource gap sum of the target field, wherein the recommendation efficiency of the target field is positively correlated with the target resource sum, and the recommendation efficiency of the target field is negatively correlated with the resource gap sum.

In the process, the recommendation efficiency of the field is determined based on the sum of target resources and the sum of resource gaps of the whole field, so that the recommendation efficiency of one field can be rapidly determined, and the judgment process and the recommendation process of the follow-up recommendation efficiency are facilitated.

According to a second aspect of embodiments of the present disclosure, there is provided a recommendation device, the device comprising:

an acquisition unit configured to perform acquisition of a first resource number of a first object, which is determined based on an object, of a plurality of objects to be recommended, whose target parameter is greater than a target threshold, and a second resource number of a second object, which is any object in a target object set, the first resource number being a resource number obtained when the first object is recommended, the second resource number being a resource number obtained when the second object is recommended;

a determining unit configured to perform determining a recommendation efficiency of the second object based on a resource gap between the first resource amount and the second resource amount, the recommendation efficiency being positively correlated with the second resource amount, the recommendation efficiency being negatively correlated with the resource gap;

and an increasing unit configured to increase the target parameter of the second object in response to the recommended efficiency of the second object reaching a first threshold.

In some embodiments, the apparatus further comprises:

a selecting unit configured to execute target parameter improvement based on the second object, and select an object with a target parameter greater than the target threshold value from the second object and the plurality of objects to be recommended;

And a recommending unit configured to perform recommending the object whose target parameter is greater than the target threshold, and record the amount of resources consumed when recommending the object whose target parameter is greater than the target threshold.

In some embodiments, the boost unit is configured to perform:

In some embodiments, the apparatus further comprises:

the dividing unit is configured to perform uniform division on the total number of resources in a target time period according to a plurality of sub-time periods in the target time period, so as to obtain the number of resources corresponding to each sub-time period, so that the total number of the resources in the plurality of sub-time periods is equal to the total number of the resources, and the total number of the resources represents the total number of the resources consumed by the recommended object in the target time period;

and the adjusting unit is configured to execute adjustment on the first threshold value based on the difference value between the number of resources corresponding to the sub-time period and the number of resources consumed by the sub-time period for any one of the plurality of sub-time periods, so as to obtain the adjusted first threshold value, wherein the first threshold value is inversely related to the difference value.

In some embodiments, the adjustment unit is further configured to perform:

In some embodiments, the apparatus further comprises:

a selection unit configured to perform selecting, in the target object set, a third resource amount of at least one third object belonging to a target domain, the third resource amount being an amount of resources obtained when recommending the third object;

the determining unit is further configured to perform determining a recommendation efficiency of the target area based on the first resource amount and a third resource amount of the at least one third object;

the improving unit is further configured to perform improving the target parameter of at least one third object included in the target area if the recommended efficiency of the target area reaches the first threshold.

In some embodiments, the determining unit comprises:

a target resource sum determination subunit configured to perform determining a target resource sum for the target domain based on the third number of resources of the at least one third object;

a resource gap sum determination subunit configured to perform determining a resource gap sum of the target area based on the resource gap between the first resource amount and a third resource amount of the at least one third object;

and a recommendation efficiency determining subunit configured to perform determining a recommendation efficiency of the target domain based on a target resource sum and a resource gap sum of the target domain, the recommendation efficiency of the target domain being positively correlated with the target resource sum, the recommendation efficiency of the target domain being negatively correlated with the resource gap sum.

According to a third aspect of embodiments of the present disclosure, there is provided a server comprising:

one or more processors;

a memory for storing the processor-executable program code;

wherein the processor is configured to execute the program code to implement the recommendation method described above.

According to a fourth aspect of embodiments of the present disclosure, there is provided a computer-readable storage medium comprising: the program code in the computer readable storage medium, when executed by a processor of a server, enables the server to perform the recommendation method described above.

According to a fifth aspect of embodiments of the present disclosure, there is provided a computer program product comprising a computer program which, when executed by a processor, implements the above-mentioned recommendation method.

It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the disclosure.

Drawings

The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the disclosure and together with the description, serve to explain the principles of the disclosure and do not constitute an undue limitation on the disclosure.

FIG. 1 is a schematic diagram of an implementation environment of a recommendation method, according to an exemplary embodiment;

FIG. 2 is a flow chart illustrating a recommendation method according to an exemplary embodiment;

FIG. 3 is a flowchart illustrating a recommendation method according to an exemplary embodiment;

FIG. 4 is a flowchart illustrating a recommendation method, according to an exemplary embodiment;

FIG. 5 is a flowchart illustrating a recommendation method, according to an exemplary embodiment;

FIG. 6 is a block diagram of a recommender arrangement in accordance with an exemplary embodiment;

fig. 7 is a block diagram of a server, according to an example embodiment.

Detailed Description

In order to enable those skilled in the art to better understand the technical solutions of the present disclosure, the technical solutions of the embodiments of the present disclosure will be clearly and completely described below with reference to the accompanying drawings.

It should be noted that the terms "first," "second," and the like in the description and claims of the present disclosure and in the foregoing 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 disclosure described herein may be capable of operation in sequences other than those illustrated or described herein. The implementations described in the following exemplary examples are not representative of all implementations consistent with the present disclosure. Rather, they are merely examples of apparatus and methods consistent with some aspects of the present disclosure as detailed in the accompanying claims.

The data referred to in this disclosure may be data authorized by the user or sufficiently authorized by the parties.

Fig. 1 is a schematic view of an implementation environment of a recommendation method provided in an embodiment of the disclosure, referring to fig. 1, where the implementation environment includes: a terminal 101 and a server 102.

The terminal 101 may be at least one of a smart phone, a smart watch, a desktop computer, a portable computer, a virtual reality terminal, an augmented reality terminal, a wireless terminal, a laptop portable computer, etc., the terminal 101 has a communication function, may access the internet, and the terminal 101 may refer to one of a plurality of terminals, which is only exemplified by the terminal 101 in this embodiment. Those skilled in the art will recognize that the number of terminals may be greater or lesser.

In some embodiments, the terminal 101 may be running a video application, a live application, a social application, or the like. During use of the video application, live application, or social application by a user, server 102 may push business objects such as advertising videos, promotional videos, etc. to these applications.

The server 102 may be an independent physical server, a server cluster or a distributed file system formed by a plurality of physical servers, or a cloud server providing cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, a content delivery network (Content Delivery Network, CDN), basic cloud computing services such as big data and an artificial intelligence platform. The server 102 may be a background server of the application programs referred to above. The server 102 and the terminal 101 may be directly or indirectly connected through wired or wireless communication, which is not limited by the embodiments of the present disclosure. In some embodiments, the number of servers 102 may be greater or lesser, as the embodiments of the present disclosure are not limited in this respect. Of course, the server 102 may also include other functional servers to provide more comprehensive and diverse services.

In the embodiment of the present disclosure, the server 102 is configured to obtain a first resource amount of a first object and a second resource amount of a second object, determine a recommendation efficiency of the second object based on a resource gap between the first resource amount and the second resource amount, and increase a target parameter of the second object in response to the recommendation efficiency of the second object reaching a first threshold. The first object is determined based on an object with a target parameter larger than a target threshold value in a plurality of objects to be recommended, the second object is any object in a target object set, the first resource quantity is the resource quantity obtained when the first object is recommended, the second resource quantity is the resource quantity obtained when the second object is recommended, the recommendation efficiency is positively correlated with the second resource quantity, and the recommendation efficiency is negatively correlated with the resource gap.

The method for allocating the denomination data can be applied to business scenes such as internet advertisements and internet propaganda, for example, can be applied to recommended scenes of business objects such as advertisement videos and propaganda videos. It should be noted that, when recommending a business object, the server 102 generally needs to go through three stages of recall, coarse ranking, and fine ranking. The recall stage is a first recommended stage, and is to quickly retrieve a part of objects potentially interested by the user from a massive object library according to the characteristics of the user and the characteristics of the object, and add the part of objects to a recall queue. The coarse ranking stage is a recommended second stage, which is to sort the objects in the recall queue according to some business indexes (such as click rate) of the coarse ranking, screen out the objects with the top sorting, and add the screened objects into the coarse ranking queue so as to reduce the number of the recalled objects and the sorting pressure of the fine ranking stage. The fine ranking stage is a third recommending stage, and objects in the coarse ranking queue are further ranked according to some fine ranking business indexes (such as thousands of times of revenues) to obtain a fine ranking queue, and the object with the highest ranking is screened out for recommendation. The disclosed embodiments subsequently employ the target parameters to represent parameters of the business index at the fine-pitch stage.

The method for allocating the denomination data provided by the embodiment of the disclosure can be applied to a fine ranking stage to recommend a part of objects preferentially. The object refers to an object belonging to a specific field or an innovative field, that is, an object containing specific content or innovative content, such as public welfare propaganda video, innovative advertisement video, and the like. It should be noted that, in the context of advertisement service or advertisement service, there is a need to preferentially recommend objects in some specific fields or innovative fields to achieve the purpose of advertising the specific fields, that is, in the fine-ranking stage, the target parameters of the objects are improved, so as to improve the opportunity of recommending the objects, thereby achieving the purpose of supporting the specific fields or innovative fields.

Fig. 2 is a flowchart of a recommendation method according to an exemplary embodiment, as shown in fig. 2, in which a server is used as an execution body, and the method includes the following steps:

in step 201, the server obtains a first resource number of a first object, which is determined based on an object, of the plurality of objects to be recommended, whose target parameter is greater than the target threshold, and a second resource number of a second object, which is any object in the target object set, the first resource number being a resource number obtained when the first object is recommended, the second resource number being a resource number obtained when the second object is recommended.

In step 202, the server determines a recommendation efficiency of the second object based on a resource gap between the first resource amount and the second resource amount, the recommendation efficiency being positively correlated with the second resource amount and the recommendation efficiency being negatively correlated with the resource gap.

In step 203, the server increases the target parameter of the second object in response to the recommended efficiency of the second object reaching the first threshold.

According to the technical scheme provided by the embodiment of the disclosure, the recommending efficiency of the object can be determined according to the number of the resources of the object with the target parameter larger than the target threshold and the number of the resources of any object in the target object set, and as the recommending efficiency is positively related to the number of the resources obtained when recommending the object, the greater the recommending efficiency of the object is, the greater the number of the resources obtained when recommending the object is, and further whether the target parameter of the object is improved is judged by judging whether the recommending efficiency reaches the first threshold or not, the object with higher recommending efficiency can be determined, more resources can be obtained, and the recommending efficiency and the maximization of the obtained number of the resources are realized.

The foregoing fig. 2 is merely a basic flow of the disclosure, and the scheme provided in the disclosure is further described below based on a specific embodiment, and fig. 3 is a flowchart of a recommendation method according to an exemplary embodiment, and referring to fig. 3, the method includes:

in step 301, the server obtains a first resource number of a first object, which is determined based on an object, of the plurality of objects to be recommended, whose target parameter is greater than the target threshold, and a second resource number of a second object, which is any object in the target object set, the first resource number being a resource number obtained when the first object is recommended, the second resource number being a resource number obtained when the second object is recommended.

The plurality of objects to be recommended may be a plurality of objects in a fine-ranking queue determined by the fine-ranking stage. The fine queuing is used for storing a plurality of objects to be recommended, and the embodiments of the present disclosure will be described below with the fine queuing as an example. The target parameter is an index parameter for sorting in the fine-ranking stage. Optionally, the target parameter may be any one of click rate, bid information, user interaction frequency, and user interest score, or may be a composite score of any two or more of the above parameters, and of course, the target parameter may also be other parameters capable of indicating that a forward effect is generated when the object is displayed. The disclosed embodiments are not limited to the target parameters selected.

The target threshold is a fixed threshold set in advance. In the embodiment of the disclosure, the object with the target parameter greater than the target threshold is used to represent the object that currently obtains the recommendation opportunity in the plurality of objects to be recommended, that is, if other objects are not considered to be recommended preferentially, the object with the target parameter greater than the target threshold is the object that will be recommended in the plurality of objects to be recommended.

The target object set refers to an object set meeting a priority recommendation condition in the objects with target parameters smaller than or equal to a target threshold value in the fine-ranking queue, namely, an object set needing priority recommendation in the objects with target parameters smaller than or equal to the target threshold value. The object satisfying the priority recommendation condition refers to an object containing specific content or innovative content, such as public welfare propaganda video, innovative advertisement video and the like. The first amount of resources is the amount of resources obtained when recommending the first object, and can be understood as the forward revenue obtained when recommending the first object. The second resource amount is the amount of resources obtained when recommending the second object, and can be understood as the forward benefit obtained when recommending the second object. Alternatively, the number of resources may be thousands of impressions (CPMs), which refers to the revenue that can be obtained per thousand impressions.

In some embodiments, the server, in response to receiving the coarse queue determined in the coarse queue stage, ranks the plurality of objects in the coarse queue according to target parameters to obtain a fine queue, and determines the first object based on the objects in the fine queue having target parameters greater than a target threshold. And determining an object set meeting the priority recommendation condition as a target object set from a plurality of objects with target parameters smaller than or equal to the target threshold in the fine-ranking queue, and selecting one object from the target object set as the second object. And respectively acquiring the first resource quantity of the first object and the second resource quantity of the second object, and then carrying out subsequent operation.

In some embodiments, the process of selecting the first object by the server includes any one of:

(1) If the target parameter of one object in the fine-queuing queue is greater than the target threshold, the server takes the object as a first object.

(2) If the target parameters of the plurality of objects in the refined queue are larger than the target threshold, the server randomly selects one object from the plurality of objects as a first object, or the server selects the object with the highest target parameter from the plurality of objects as the first object. In the process, if a plurality of objects with target parameters larger than the target threshold exist, one of the objects or the object with the highest target parameter can be selected as the first object, so that the first object can be quickly determined, and the subsequent operation can be conveniently executed.

(3) And the server selects an object with the highest target parameter from the objects included in the fine-ranking queue as a first object. That is, the server acquires the object with the front ranking order from the fine ranking queue as the first object. Therefore, the first object can be determined more quickly by directly selecting the object with the front arrangement sequence in the queue, the efficiency of determining the first object is improved, and the recommendation efficiency is further improved. The embodiment of the disclosure takes the object with the highest target parameter as the first object as an example.

In some embodiments, the process of selecting the second object by the server includes any one of:

(1) The server randomly selects an object from the target object set as the second object. In the embodiment, the second object is selected in a random selection mode, so that the second object can be rapidly determined, and the recommendation efficiency is improved.

(2) And the server selects an object with the highest target parameter from the target object set as the second object. In this embodiment, the object with the highest target parameter is selected, so that the second object can be determined quickly, and the possibility of preferentially recommending the second object can be improved.

In some embodiments, taking thousands of revenues as an example, the process of obtaining, by the server, the first resource amount of the first object and the second resource amount of the second object is: predicting click rates of a first object and a second object to obtain predicted click rates of the first object and the second object, determining the product of the predicted click rate, single click profit and a constant 1000 of the first object, displaying the profit for thousands of times of the first object, taking the display profit for thousands of times of the first object as the first resource quantity of the first object, determining the product of the predicted click rate, single click profit and the constant 1000 of the second object, displaying the profit for thousands of times of the second object, and taking the display profit for thousands of times of the second object as the second resource quantity of the second object. In the embodiment, the resource quantity of the corresponding object can be rapidly determined by predicting thousands of display benefits, so that the process of recommending the subsequent object is facilitated.

The above-mentioned process is a process of acquiring the first resource quantity of the first object and the second resource quantity of the second object after determining the first object and the second object based on the target parameter. In other embodiments, the target parameters and the number of resources may each exhibit revenue for thousands of times. Taking the object with the highest target parameter as the first object as an example, the corresponding procedure in step 301 is as follows: and the server responds to the received coarse queuing and determines the first resource quantity of the first object, wherein the coarse queuing is determined in the coarse queuing stage, the first resource quantity is obtained by sequencing thousands of display benefits of a plurality of objects in the coarse queuing to obtain a fine queuing, and thousands of display benefits of the objects with the forefront sequencing in the fine queuing are determined. And determining an object set meeting the priority recommendation condition as a target object set in a plurality of objects except the first object in the fine-ranking queue, selecting thousands of display benefits of one object from the target object set as a second resource quantity of the second object, and then carrying out subsequent operation.

In step 302, the server determines a recommendation efficiency of the second object based on a resource gap between the first resource amount and the second resource amount, the recommendation efficiency being positively correlated with the second resource amount and the recommendation efficiency being negatively correlated with the resource gap.

The resource gap may be a resource difference between the first resource number and the second resource number, and the description will be given by taking the resource difference as an example. The resource gap is used to represent the amount of resources consumed in recommending the second object, and can be understood as a negative loss in recommending the second object. For example, if the first number of resources is 10 and the second number of resources is 5, the number of resources obtained when the first object is recommended is 10, the number of resources obtained when the second object is recommended is 5, and for preferentially recommending the second object, the negative loss is the difference (10-5=5) between the first number of resources and the second number of resources. It should be understood that a negative loss is a subsidy of the number of resources, so that the process of preferentially recommending can also be understood as a subsidy process of the number of resources, and the recommending efficiency can also be understood as a subsidy use efficiency.

It should be noted that, the plurality of objects in the target object set are all the objects that are not the first object in the order of step 301, so the number of resources obtained by the plurality of objects in the target object set is lower than the number of resources obtained by the first object (the object that is the first object in the order), so the difference between the resources between any object in the target object set and the first object is the number of resources obtained by the any object is lower than the number of resources obtained by the first object. It should be noted that, the first resource amount, the second resource amount, and the resource difference between the first resource amount and the second resource amount are all predicted in the fine-ranking stage.

In some embodiments, after the server obtains the first number of resources of the first object and the second number of resources of the second object, a recommendation efficiency for the second object is determined based on a resource gap between the first number of resources and the second number of resources, and equation (1).

Where bonus_roi is the recommended efficiency of the second object, cpm_reduce_origin is the second resource quantity of the second object, cpm_left is the first resource quantity of the first object, and cpm_reduce_origin-cpm_left is the resource gap between the first resource quantity and the second resource quantity. In the process, the recommendation efficiency is defined, and the larger the second resource quantity of the second object is, the smaller the resource gap between the first resource quantity and the second resource quantity is, the larger the recommendation efficiency is, and the second object is preferably recommended, so that the obtained positive benefit can be maximized under the condition that the negative loss is minimum.

In some embodiments, the server calculates the recommendation efficiency of the second object by using the above formula (1) and then performs the following steps in case that the second object has a possibility of being recommended preferentially and has a negative effect. To facilitate an understanding of this process, the following is a detailed description based on three possible scenarios that may exist:

Firstly, under the condition that priority recommendation is not considered, the first resource quantity of an object (first object) with the highest target parameter is cpm_left (follow-up abbreviated as non-subsidized top1 benefit); in the scenario of considering priority recommendation, the resource number of the object with the highest target parameter is cpm_reduction_bonus (hereinafter referred to as subsidized top1 benefit for short), and it should be noted that, at this time, the object with the highest target parameter may be the first object or the second object after parameter reordering is improved; the second resource amount of the second object is cpm_yield_origin (hereinafter abbreviated as the benefit of the second object). Then, when recommending an object, there may be three cases shown in fig. 1.

TABLE 1

As shown in table 3, the three cases are respectively:

(1) The benefit of the subsidy top1 > the benefit of the second object > the benefit of the non-subsidy top1, in this case, the benefit of the subsidy top1 is the largest, and the object with the largest benefit is the second object with the improved parameter reordered, so that the second object has the possibility of being recommended preferentially, and the benefit of the second object is greater than the benefit of the non-subsidy top1, namely, the benefit of the second object is greater than the benefit of the first object, in this case, if the second object is recommended preferentially, negative influence is not generated;

(2) The benefit of the subsidy top1 > the benefit of the non-subsidy top1 > the benefit of the second object, in this case, the benefit of the subsidy top1 is the largest, and the object with the largest benefit is the second object with the improved parameter reordered, so that the second object has the possibility of being recommended preferentially, the benefit of the non-subsidy top1 is greater than the benefit of the second object, that is, the benefit of the second object is less than the benefit of the first object, in this case, if the second object is recommended preferentially, negative influence can be generated;

(3) The benefit of the patch-free top1 > the benefit of the second object, in which case the benefit of the patch-free top1 is smaller than the benefit of the patch-free top1, so that there is no possibility of preferential recommendation, and negative effects are not generated.

Thus, as can be seen from the above discussion, when there is a possibility of priority recommendation and the second resource amount of the second object is smaller than the first resource amount of the first object, the above formula (1) can be used to calculate the recommendation efficiency of the second object, and the corresponding procedure can be: before executing step 301, the target parameters of the second object may be increased in advance, reordered according to the target parameters after the second object is increased, and then judged according to the target parameters after the second object is increased, the original target parameters of the second object, and the target parameters of the first object, to determine whether the second object belongs to the case (2), that is, whether the second object belongs to the case where there is a possibility of being preferentially recommended and a negative influence is generated, if the second object belongs to the case where there is a possibility of being preferentially recommended and a negative influence is generated, the recommendation efficiency of the second object is calculated by adopting the above formula (1), and if the second object does not belong to the case where there is a possibility of being preferentially recommended and a negative influence is generated, the recommendation efficiency of the second object does not need to be calculated. Therefore, for some objects which are not likely to be recommended preferentially or do not have negative influence, calculation of recommendation efficiency is not needed, processing content of the server is greatly reduced, processing speed of the server is improved, recommendation efficiency of the second object can be effectively determined, and accuracy of the recommended objects is improved.

In step 303, the server increases the target parameter of the second object in response to the recommended efficiency of the second object reaching the first threshold.

The first threshold is a preset fixed threshold.

In some embodiments, the server increases the target parameter of the second object according to a preset adjustment value in response to the recommended efficiency of the second object reaching the first threshold. For example, if the target parameter of the second object is 5 and the adjustment value is 2, the target parameter after the improvement is 5*2 =10.

In some embodiments, after determining the recommendation efficiency of the second object, the server increases the target parameter of the second object in response to the recommendation efficiency of the second object reaching a first threshold and the amount of resources consumed by the currently recommended object satisfying the target condition. In the process, when judging whether to improve the target parameters of the second object, not only the recommendation efficiency and the first threshold value but also the amount of resources consumed by the recommended object are considered, so that the excessive amount of resources consumed is avoided.

In other embodiments, the target parameter of the second object is not increased in response to the recommended efficiency of the second object not reaching the first threshold.

Wherein the target condition is a preset condition. Optionally, the target condition is that the amount of resources consumed by the currently recommended object is less than the total amount of resources, which is an upper limit of the amount of resources consumed. It should be understood that in a scenario where there is no priority recommendation, the first object is the original recommended object, the amount of resources obtained when the server recommends the first object is usually the largest compared to other objects, in a scenario where some objects are preferentially recommended, the recommendation of the original recommended object (i.e. the first object) needs to be canceled, and the amount of resources obtained when the preferentially recommended object is recommended is usually smaller than the amount of resources of the first object, so that, in order to avoid excessive negative loss, the threshold of negative loss is usually limited, that is, the total amount of resources mentioned above.

The above-mentioned amount of resources consumed by the currently recommended object refers to the amount of resources consumed when the object is actually recommended. It should be appreciated that prior to implementing the present solution, the server may set the total number of resources for a target time period (e.g., one day) to limit the amount of resources consumed to not exceed the set total number of resources, which represents the total number of resources consumed to recommend the subject during the target time period. For example, the total number of resources may be 50, and in response to the recommended efficiency of the second object reaching the first threshold, the determining is performed according to the number of resources consumed by the currently recommended object and the total number of resources 50, if the number of resources consumed by the currently recommended object is less than 50 (e.g. 40), the step of increasing the target parameter of the second object is performed, and if the number of resources consumed by the currently recommended object is greater than or equal to 50 (e.g. 55), the step of increasing the target parameter of the second object is not performed.

The above-mentioned process is a process of judging whether the number of resources consumed by the currently recommended object satisfies the target condition or not when the recommending efficiency of the second object reaches the first threshold. In other embodiments, the server can also determine whether the number of resources consumed by the currently recommended object meets the target condition, and execute step 301 and step 302 when the number of resources consumed by the currently recommended object meets the target condition. Therefore, whether the consumed resource quantity meets the target condition can be known in advance, the server continues to determine the recommendation efficiency under the condition that the consumed resource quantity meets the target condition, and the process of determining the recommendation efficiency is not required to be executed under the condition that the consumed resource quantity does not meet the target condition, so that the processing content of the server is greatly reduced. For example, the total number of resources may be 50, if the number of resources consumed by the currently recommended object is less than 50 (e.g. 40), then steps 301 and 302 are continued, and if the number of resources consumed by the currently recommended object is greater than or equal to 50 (e.g. 55), then steps 301 and 302 are not further performed.

In some embodiments, the server uniformly divides the total number of resources in the target time period according to the plurality of sub-time periods in the target time period to obtain the number of resources corresponding to each sub-time period, so that the sum of the number of resources in the plurality of sub-time periods is equal to the total number of resources. Further, for any one of the plurality of sub-time periods, based on the number of resources corresponding to the sub-time period and the number of resources consumed by the sub-time period, it is determined whether the number of resources consumed by the object currently recommended for the sub-time period is smaller than the number of resources corresponding to the corresponding sub-time period, if the number of resources consumed by the object currently recommended for the sub-time period is smaller than the number of resources corresponding to the corresponding sub-time period, the process of recommending the object preferentially is continued, and if the number of resources consumed by the object currently recommended for the sub-time period is greater than or equal to the number of resources corresponding to the corresponding sub-time period, the process of recommending the object preferentially is not executed. In this embodiment, the server may further smooth the total number of resources in the target time period to each sub-time period of the target time period, so as to avoid that a large number of objects are raised by the target parameter in the initial stage of the target time period, and further, the objects in the remaining time period cannot obtain the opportunity of being raised by the target parameter, so that the speed of the process of raising the target parameter in the target time period can be effectively controlled.

Optionally, the server divides the total number of resources in the target time period based on a time granularity of an hour level, for example, 1 hour, or the server divides the total number of resources in the target time period based on a time granularity of a minute level, for example, 5 minutes. In this way, the use speed of the total number of resources can be smoothed, so that the total number of resources runs out smoothly within the target period of time.

In some embodiments, the server uniformly divides the total number of resources in the target time period according to the plurality of sub-time periods in the target time period to obtain the number of resources corresponding to each sub-time period, so that the sum of the number of resources in the plurality of sub-time periods is equal to the total number of resources. And for any one of the plurality of sub-time periods, adjusting the first threshold based on a difference value between the number of resources corresponding to the sub-time period and the number of resources consumed by the sub-time period to obtain the adjusted first threshold, wherein the first threshold is inversely related to the difference value.

Optionally, the server adjusts the first threshold based on a PID (proportional-integral-derivative) feedback regulator. Fig. 4 is a flowchart of a recommendation method according to an exemplary embodiment, referring to fig. 4, for any one of the sub-time periods, the PID feedback regulator is input with the number of resources corresponding to the sub-time period and the number of resources consumed by the sub-time period, and then the PID feedback regulator calculates the difference between the number of resources corresponding to the sub-time period and the number of resources consumed by the sub-time period, and determines a first threshold corresponding to the difference, thereby obtaining an adjusted first threshold. In the process, a PID feedback regulator is adopted, and the first threshold value is regulated, so that the total amount of resources is smoothly used up in a target time period, and a large amount of objects are prevented from consuming the total amount of resources in the initial stage of the target time period.

In this embodiment, the server not only can smooth the total number of resources in the target time period to each sub-time period of the target time period, so as to avoid that a large number of objects are raised in the initial stage of the target time period, and further, the objects in the remaining time period cannot be raised in the target parameter, but also can effectively control the speed of raising the target parameter in the target time period, and can adjust the first threshold based on the real-time consumed number of resources, when the consumed number of resources is small and the difference value is large, the first threshold is lowered, the target parameter of the second object is raised as much as possible, when the consumed number of resources is large and the difference value is small, the first threshold is raised, the opportunity that the second object is raised in the target parameter is lowered, the excessive consumed number of resources is avoided, and the optimization is performed under the condition that the total number of resources is ensured to be smoothly utilized in the target time period, and the recommended efficiency is maximized.

In other embodiments, the server obtains, from the plurality of recommended objects in the target time period, a consumed resource amount of the plurality of objects belonging to the target publisher, increases the first threshold of the plurality of objects belonging to the target publisher if the consumed resource amount is greater than the second threshold, and decreases the first threshold of the plurality of objects belonging to the target publisher if the consumed resource amount is less than or equal to the second threshold. Wherein the plurality of objects belonging to the target publisher refer to a plurality of objects belonging to the same advertiser. In this embodiment, the first threshold may be appropriately raised for publishers with larger total consumption of resources, and the first threshold may be appropriately lowered for publishers with smaller total consumption of resources, thereby improving fairness and health of priority recommendation.

It should be noted that in the above process, a situation that patches are concentrated on some head products in a large amount may occur, in order to balance the opportunities of improving the target parameters, to ensure a certain fairness, the first threshold may be adjusted based on the total consumption of each publisher, so as to improve fairness and health of priority recommendation.

In step 304, the server selects, based on the target parameter after the second object is improved, an object whose target parameter is greater than a target threshold value from the second object and the plurality of objects to be recommended.

In some embodiments, after the server increases the target parameter of the second object, the server reorders the second object and the plurality of objects to be recommended based on the target parameter after the second object is increased, selects an object whose target parameter is greater than the target threshold, and then performs the subsequent recommendation operation.

Optionally, after the server increases the target parameters of the second object, based on the target parameters after the second object is increased, the second object and the plurality of objects to be recommended are reordered, and the object with the highest target parameters is selected, that is, the object with the highest target parameters is selected, and then the subsequent recommendation operation is performed.

The steps 301 to 304 are processes of determining whether to raise the target parameter of the second object based on the second object in the target object set, and then redetermining the object whose target parameter is greater than the target threshold. It should be noted that after determining whether to increase the target parameter of the second object, the server needs to continue to traverse the rest objects in the target object set until the number of resources consumed by the currently recommended objects reaches the total number of resources, and stopping the process of determining the recommendation efficiency and determining whether the recommendation efficiency is greater than the first threshold. The corresponding process is as follows: after the server improves the target parameters of the second object, continuing to select the next object in the target object set, acquiring the resource quantity of the next object, determining the recommending efficiency of the next object based on the resource quantity of the next object and the first resource quantity of the first object, improving the target parameters of the next object in response to the recommending efficiency of the next object reaching a first threshold, repeating the processes of selecting the next object and determining the recommending efficiency of the next object until the resource quantity consumed by the currently recommended object reaches the total resource quantity, and stopping executing the processes of selecting the next object and determining the recommending efficiency of the next object.

It should be further noted that, step 304 may be performed after determining a plurality of objects for improving the target parameters in the target object set, that is, after determining a plurality of objects for which the recommendation efficiency reaches the first threshold, the server improves the target parameters of the plurality of objects, reorders the plurality of objects and the plurality of objects to be recommended based on the improved target parameters of the plurality of objects, and selects an object with the highest ranking, that is, selects an object with the target parameter greater than the target threshold. In the process, the reordering process is performed without determining an object with the recommended efficiency reaching the first threshold value, so that the processing content of the server is greatly reduced, and the processing speed of the server is improved.

In step 305, the server recommends objects whose target parameters are greater than the target threshold and records the amount of resources consumed in recommending objects whose target parameters are greater than the target threshold.

In some embodiments, after determining that the target parameter is greater than the target threshold, the server recommends the object whose target parameter is greater than the target threshold, obtains the number of resources obtained when recommending the object whose target parameter is greater than the target threshold, determines a difference between the first number of resources of the first object and the number of resources obtained when recommending the object whose target parameter is greater than the target threshold, and records the number of resources consumed when recommending the object whose target parameter is greater than the target threshold, thereby facilitating execution of a subsequent recommendation process. In the process, according to the target parameters of the second object after being improved and the target parameters of the plurality of objects to be recommended, selecting the objects with the target parameters larger than the target threshold value for recommendation, and when the priority recommendation of some objects is realized, the recommendation effect of the objects can be ensured, and the quantity of resources obtained during the recommendation of the objects is improved.

FIG. 3 is a process of determining whether to increase the target parameters of the second object based on the recommendation efficiency of any object in the target object set, and then performing recommendation. In other embodiments, the server can also determine the recommendation efficiency of a domain based on a plurality of objects in the domain, further determine whether to increase the target parameters of the plurality of objects included in the domain, and perform the recommendation process. FIG. 5 is a flowchart illustrating a recommendation method, see FIG. 5, according to an exemplary embodiment, the method comprising:

In step 501, the server obtains a first resource amount of a first object, where the first object is determined based on an object, of the plurality of objects to be recommended, whose target parameter is greater than a target threshold, and the first resource amount is a resource amount obtained when the first object is recommended.

It should be noted that, the content of step 501 is referred to step 301, and will not be described again.

In step 502, the server selects a third resource amount of at least one third object belonging to the target domain from the target object set, where the third resource amount is an amount of resources obtained when recommending the third object.

The target domain refers to a service domain satisfying a priority recommendation condition, for example, a specific domain or an innovation domain. The field is understood to be an industry, then the target field may be a particular industry or an innovative industry. Accordingly, objects belonging to the target field are objects belonging to a specific industry or an innovative industry, such as public welfare videos and innovative videos. It should be understood that each object carries an identification for indicating the field. The third resource amount is the amount of resources obtained when recommending the third object, and can be understood as the forward benefit obtained when recommending the third object.

In some embodiments, the server selects at least one third object belonging to the target domain according to the identifiers of the plurality of objects in the target object set, and obtains a third resource amount of the at least one third object. In the process, the number of resources of the objects belonging to the same field is selected, so that the follow-up process of determining the recommendation efficiency of the field is facilitated.

In step 503, the server determines a recommendation efficiency for the target area based on the first number of resources and the third number of resources of the at least one third object.

In some embodiments, the server determines a target resource sum of the target domain based on the third resource amount of the at least one third object, determines a resource gap sum of the target domain based on a resource gap between the first resource amount and the third resource amount of the at least one third object, determines a recommendation efficiency of the target domain based on the target resource sum and the resource gap sum of the target domain, the recommendation efficiency of the target domain being positively correlated with the target resource sum, and the recommendation efficiency of the target domain being negatively correlated with the resource gap sum.

In some embodiments, the process of determining the recommended efficiency of the target domain by the server based on the target resource sum and the resource gap sum of the target domain includes: the server determines the recommendation efficiency of the target domain based on the target resource sum, the resource gap sum and the formula (2) of the target domain.

In the formula, bonus_roi' is the recommended efficiency of the target field, sigma cpm_index_origin is the target resource sum of the target field, and sigma|cpm_index_cpm_left| is the resource gap sum of the target field.

In the process, the recommendation efficiency of the field is determined based on the sum of target resources and the sum of resource gaps of the whole field, so that the recommendation efficiency of one field can be rapidly determined, and the subsequent recommendation efficiency judging process and the recommendation process are facilitated.

In step 504, the server increases a target parameter of at least one third object included in the target area in response to the recommended efficiency of the target area reaching the first threshold.

In the process, the number of the resources of the plurality of objects belonging to the same field is selected, so that the recommendation efficiency of the field is determined, whether the target parameters of the plurality of objects are improved can be judged more quickly, the recommendation efficiency is not required to be determined for the objects one by one, and the execution efficiency of the whole recommendation process is improved.

It should be noted that, after determining whether to increase the target parameter of at least one third object included in the target domain, the server needs to further traverse the objects in the other domains in the target object set until the number of resources consumed by the currently recommended object reaches the total number of resources, and stopping the process of determining the recommendation efficiency and determining whether the recommendation efficiency is greater than the first threshold. The corresponding process is as follows: after the server improves the target parameters of at least one third object included in the target domain, continuing to select an object belonging to another domain from the target object set, acquiring the resource quantity of the object belonging to another domain, determining the recommendation efficiency of the another domain based on the resource quantity of the object of the another domain and the first resource quantity of the first object, improving the target parameters of the object of the another domain in response to the recommendation efficiency of the another domain reaching a first threshold, repeatedly executing the process of selecting the another domain and determining the recommendation efficiency until the resource quantity consumed by the currently recommended object reaches the total resource quantity, and stopping executing the process of selecting the another domain and determining the recommendation efficiency.

In step 505, the server selects, based on the target parameter raised by the at least one third object, an object whose target parameter is greater than a target threshold value from the at least one third object and the plurality of objects to be recommended.

In step 506, the server recommends objects whose target parameters are greater than the target threshold, and records the amount of resources consumed in recommending objects whose target parameters are greater than the target threshold.

The contents of step 504 and step 506 refer to step 303 to step 305, and are not described in detail.

FIG. 6 is a block diagram illustrating a recommender in accordance with an exemplary embodiment. Referring to fig. 6, the apparatus includes an acquisition unit 601, a determination unit 602, and an improvement unit 603.

An obtaining unit 601 configured to perform obtaining a first resource amount of a first object and a second resource amount of a second object, where the first object is an object whose target parameter is greater than a target threshold value among a plurality of objects to be recommended, the second object is any object in a target object set, the first resource amount is a resource amount obtained when the first object is recommended, and the second resource amount is a resource amount obtained when the second object is recommended;

a determining unit 602 configured to perform determining a recommendation efficiency of the second object based on a resource gap between the first resource amount and the second resource amount, the recommendation efficiency being positively correlated with the second resource amount, the recommendation efficiency being negatively correlated with the resource gap;

an increasing unit 603 is configured to perform increasing the target parameter of the second object in response to the recommended efficiency of the second object reaching a first threshold.

In some embodiments, the apparatus further comprises:

a selecting unit configured to execute target parameter improvement based on the second object, and select an object with a target parameter greater than a target threshold value from the second object and the plurality of objects to be recommended;

And a recommending unit configured to perform recommending the object whose target parameter is greater than the target threshold value, and record the amount of resources consumed when recommending the object whose target parameter is greater than the target threshold value.

In some embodiments, the boosting unit 603 is configured to perform:

In some embodiments, the apparatus further comprises:

In some embodiments, the adjustment unit is further configured to perform:

In some embodiments, the apparatus further comprises:

the determining unit 602 is further configured to perform determining a recommendation efficiency for the target area based on the first number of resources and the third number of resources of the at least one third object;

the improving unit 603 is further configured to perform improving the target parameter of the at least one third object included in the target area if the recommended efficiency of the target area reaches the first threshold.

In some embodiments, the determining unit 602 includes:

It should be noted that: in the recommendation device provided in the above embodiment, only the division of the above functional modules is used for illustration, and in practical application, the above functional allocation may be performed by different functional modules according to needs, that is, the internal structure of the device is divided into different functional modules, so as to complete all or part of the functions described above. In addition, the recommending device and the recommending method provided in the above embodiments belong to the same concept, and the specific implementation process is detailed in the method embodiment, which is not described herein again.

Fig. 7 is a block diagram illustrating a server 700, which may be configured or perform differently to a large extent, may include one or more processors (Central Processing Units, CPU) 701 and one or more memories 702, wherein the one or more memories 702 store at least one program code that is loaded and executed by the one or more processors 701 to implement the recommended methods provided by the various method embodiments described above, according to an exemplary embodiment. Of course, the server 700 may also have a wired or wireless network interface, a keyboard, an input/output interface, and other components for implementing the functions of the device, which are not described herein.

In an exemplary embodiment, a computer readable storage medium is also provided, comprising program code, for example a memory 702 comprising program code, which is executable by the processor 701 of the server 700 to perform the above-mentioned recommendation method. Alternatively, the computer readable storage medium may be a read-only memory (ROM), a random access memory (random access memory), a RAM), a compact-disk read-only memory (CD-ROM), a magnetic tape, a floppy disk, an optical data storage device, and the like.

In an exemplary embodiment, a computer program product is also provided, comprising a computer program which, when executed by a processor, implements the above-mentioned recommendation method.

Other embodiments of the disclosure will be apparent to those skilled in the art from consideration of the specification and practice of the disclosure disclosed herein. This disclosure is intended to cover any adaptations, uses, or adaptations of the disclosure following the general principles of the disclosure and including such departures from the present disclosure as come within known or customary practice within the art to which the disclosure pertains. It is intended that the specification and examples be considered as exemplary only, with a true scope and spirit of the disclosure being indicated by the following claims.

It is to be understood that the present disclosure is not limited to the precise arrangements and instrumentalities shown in the drawings, and that various modifications and changes may be effected without departing from the scope thereof. The scope of the present disclosure is limited only by the appended claims.

Claims

1. A recommendation method, the method comprising:

2. The recommendation method according to claim 1, wherein after increasing the target parameter of the second object in response to the recommendation efficiency of the second object reaching a first threshold, the method further comprises:

selecting an object with a target parameter larger than the target threshold value from the second object and the plurality of objects to be recommended based on the target parameter of the second object after the improvement;

recommending the object with the target parameter larger than the target threshold, and recording the amount of resources consumed when recommending the object with the target parameter larger than the target threshold.

3. The recommendation method of claim 1, wherein increasing the target parameter of the second object in response to the recommendation efficiency of the second object reaching a first threshold comprises:

and in response to the recommending efficiency of the second object reaches a first threshold, and the quantity of resources consumed by the currently recommended object meets a target condition, improving the target parameter of the second object.

4. The recommendation method of claim 1, wherein prior to said obtaining the first number of resources of the first object and the second number of resources of the second object, the method further comprises:

5. The recommendation method of claim 4, further comprising:

6. The recommendation method according to claim 1, wherein said method further comprises:

determining a recommendation efficiency for the target area based on the first number of resources and a third number of resources for the at least one third object;

and if the recommended efficiency of the target field reaches the first threshold, improving the target parameter of at least one third object included in the target field.

7. The recommendation method of claim 6, wherein said determining a recommendation efficiency for said target area based on said first number of resources and a third number of resources for said at least one third object comprises:

determining a target resource sum of the target domain based on a third number of resources of the at least one third object;

determining a sum of resource gaps for the target area based on the resource gaps between the first number of resources and a third number of resources of the at least one third object;

8. A recommendation device, the device comprising:

9. The recommendation device of claim 8, wherein the device further comprises:

a selecting unit configured to perform a target parameter based on the second object after the improvement, and select an object whose target parameter is greater than the target threshold value from the second object and the plurality of objects to be recommended;

And a recommending unit configured to perform recommending of the object whose target parameter is greater than the target threshold value, and to record the amount of resources consumed when recommending the object whose target parameter is greater than the target threshold value.

10. The recommendation device of claim 8, wherein the boost unit is configured to perform:

11. The recommendation device of claim 8, wherein the device further comprises:

the dividing unit is configured to perform uniform division on the total number of resources in a target time period according to a plurality of sub-time periods in the target time period, so as to obtain the number of resources corresponding to each sub-time period, so that the total number of the resources in the plurality of sub-time periods is equal to the total number of the resources, and the total number of the resources represents the total number of the resources consumed by a recommended object in the target time period;

12. The recommendation device of claim 11, wherein the adjustment unit is further configured to perform:

13. The recommendation device of claim 8, wherein the device further comprises:

a selecting unit configured to perform selecting, in the target object set, a third resource amount of at least one third object belonging to a target domain, the third resource amount being a resource amount obtained when recommending the third object;

the increasing unit is further configured to increase a target parameter of at least one third object included in the target area if the recommended efficiency of the target area reaches the first threshold.

14. The recommendation device according to claim 13, wherein said determining unit comprises:

a target resource sum determination subunit configured to perform determining a target resource sum of the target domain based on a third number of resources of the at least one third object;

a resource gap sum determination subunit configured to perform determining a resource gap sum of the target area based on a resource gap between the first resource amount and a third resource amount of the at least one third object;

and a recommendation efficiency determining subunit configured to determine a recommendation efficiency of the target domain based on a target resource sum and a resource gap sum of the target domain, the recommendation efficiency of the target domain being positively correlated with the target resource sum, the recommendation efficiency of the target domain being negatively correlated with the resource gap sum.

15. A server, the server comprising:

one or more processors;

a memory for storing the processor-executable program code;

wherein the processor is configured to execute the program code to implement the recommendation method of any one of claims 1 to 7.

16. A computer readable storage medium, characterized in that program code in the computer readable storage medium, when executed by a processor of a server, enables the server to perform the recommendation method according to any one of claims 1 to 7.