Advertisement sorting method and device, electronic equipment and storage medium

Abstract

The embodiment of the invention provides an advertisement sorting method and device, electronic equipment and a storage medium, and relates to the technical field of data processing.The method comprises the steps that in response to an advertisement exposure request, candidate advertisements are recalled; determining a first advertisement of which the advertisement feature is matched with the interest feature of a first user from the candidate advertisements, the first user being a user identified by the user identity identifier carried by the advertisement exposure request; obtaining a push weighting coefficient of the first advertisement; adjusting a pushing coefficient of the first advertisement based on the pushing weighting coefficient; and sorting the candidate advertisements on the basis of the adjusted push coefficient. When the scheme provided by the embodiment of the invention is applied to sort the advertisements, the push coefficient of the first advertisement matched with the interest feature of the first user is adjusted, so that the probability that the first advertisement is pushed preferentially can be increased, and the conversion rate of the advertisements is further improved.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to an advertisement sorting method, device, electronic device and storage medium. Background Art

[0002] In today's world of great material abundance, there is a wide variety of goods. Merchants hope that more and more users will be able to understand their products, thereby promoting their sales. In view of this, various merchants will place advertisements on various platforms in order to increase the exposure of their products and expand the audience of their products.

[0003] When a user meets the conditions for ad push, the platform recalls the ads, sorts the recalled ads, and then pushes the ads to the user according to the sorting results. It can be seen that the sorting method directly affects the ad push results. Therefore, it is necessary to provide an ad sorting solution to improve the conversion rate of ads when pushing ads to users. Summary of the invention

[0004] The purpose of the embodiments of the present invention is to provide an advertisement sorting method, device, electronic device and storage medium, which can sort advertisements that users are interested in in a personalized manner and improve conversion rates. The specific scheme is as follows:

[0005] In a first aspect, an embodiment of the present invention provides an advertisement ranking method, the method comprising:

[0006] Recalling candidate advertisements in response to the advertisement exposure request;

[0007] Determine, from the candidate advertisements, a first advertisement whose advertisement characteristics match the interest characteristics of a first user, wherein the first user is: a user identified by a user identity carried in the advertisement exposure request;

[0008] Obtaining a push weighting coefficient for the first advertisement;

[0009] Based on the push weighting coefficient, adjusting the push coefficient of the first advertisement;

[0010] The candidate advertisements are ranked based on the adjusted push coefficients.

[0011] In one embodiment of the present invention, the push weighting coefficient is determined based on the interest representation value corresponding to the first interested group, the first interested group is: a set of user identity identifiers of users who are interested in the advertisement commissioned by the first advertiser, the interest representation value represents the degree of interest of the users identified by the user identity identifiers in the first interested group in the advertisement commissioned by the first advertiser, and the first advertiser is: the advertiser of the first advertisement.

[0012] In one embodiment of the present invention, the interested groups corresponding to the advertiser are determined in the following manner:

[0013] Based on the historical behavior of the user identified by the user identity identifier, the user identity identifier is classified to obtain a group of people corresponding to users with similar interests;

[0014] The interest representation values ​​of the groups in each group-advertiser pair towards the advertiser's advertisements within the sample time period are counted to obtain the corresponding statistical results of each group-advertiser pair;

[0015] Based on the corresponding statistical results of each group-advertiser pair, the overall interest representation value of the group in each group-advertiser pair for the advertiser's advertisement is obtained;

[0016] It is determined that the population in the advertiser pair whose overall interest representation value is greater than a preset interest threshold is the advertiser's interested population.

[0017] In one embodiment of the present invention, the push weighting coefficient is determined based on the interest representation value corresponding to the first interested group under the preset conversion evaluation dimension;

[0018] The statistical analysis of the interest representation values ​​of the groups of people in each group-advertiser pair in the sample time period to the advertiser's advertisements is performed to obtain the corresponding statistical results of each group-advertiser pair, including:

[0019] Statistics are collected on the interest representation values ​​of the advertisers in each group of advertisers in the preset conversion evaluation dimension during the sample period to obtain the corresponding statistical results of the advertisers in each group of advertisers in the preset conversion evaluation dimension.

[0020] In one embodiment of the present invention, the preset conversion evaluation dimension includes at least one of the following dimensions:

[0021] The user's interest in the target category, wherein the target category is: the category to which the advertiser's advertisement belongs;

[0022] Parameters that directly describe whether a user converts an advertiser's ad;

[0023] Parameters that indirectly describe whether a user converts an advertiser's ad;

[0024] The sales revenue brought to the advertiser by the target user.

[0025] In one embodiment of the present invention, the user identity identifiers are classified based on the historical behaviors of the users identified by the user identity identifiers to obtain the groups of people corresponding to users with similar interests, including:

[0026] The user identity is classified according to at least one of the following characteristics of the historical behavior of the user identified by the user identity to obtain a group of users with similar interests:

[0027] A first feature updated according to the user's operation behavior on the played advertisement in a recent historical period, wherein the first feature represents whether the user is interested in the advertisement;

[0028] A second feature updated according to the operation behavior of the user on the advertising promotion platform in the recent historical time period, wherein the second feature represents an attribute of the user itself;

[0029] A third feature is generated based on the interest prediction model, wherein the third feature represents the interest of the user.

[0030] In one embodiment of the present invention, the push weighting coefficient is generated in the following manner:

[0031] Obtaining the number of advertisement conversions brought to the first advertiser by the user identified by the user identity identifier in the first interested group in the first time period;

[0032] Obtaining an interest representation value expressed by a user identified by a user identity identifier in the first interested group in the advertisement commissioned by the first advertiser during the first time period;

[0033] Determine, based on the obtained interest representation value, the number of advertisement conversions, and an expected value of a preset conversion indicator, a conversion contribution representation value of the user identified by the user identity in the first interested group to the first advertiser;

[0034] The push weighting coefficient is generated based on the conversion contribution representation value.

[0035] In one embodiment of the present invention, determining the conversion contribution representation value of the user identified by the user identity in the first interested group to the first advertiser based on the obtained interest representation value, the number of advertisement conversions, and the expected value of the preset conversion indicator includes:

[0036] Normalizing the obtained interest representation value and the expected value of the preset conversion indicator;

[0037] If the number of advertisement conversions is greater than a preset number threshold, an adjustment value is generated based on a preset adjustment coefficient and a normalized expected value, and the normalized interest representation value is adjusted based on the adjustment value to obtain a conversion contribution representation value of the user identified by the user identity identifier in the first interested group to the first advertiser;

[0038] If the number of advertisement conversions is not greater than the preset threshold value, an adjustment value is generated based on a preset adjustment coefficient and a normalized expected value as a conversion contribution representation value.

[0039] In one embodiment of the present invention, generating the push weighting coefficient based on the conversion contribution representation value includes:

[0040] If the conversion contribution representation value is greater than a preset contribution threshold, mapping transformation is performed on the conversion contribution representation value according to a preset mapping rule, and the minimum value between the mapping transformation result and a preset maximum push weighting coefficient is determined as the push weighting coefficient;

[0041] If the conversion contribution representation value is not greater than the preset contribution threshold and is greater than 0, determining the push weighting coefficient to be a first value;

[0042] If the conversion contribution representation value is less than or equal to 0, the push weighting coefficient is determined to be a second value, wherein the second value is less than the first value.

[0043] In one embodiment of the present invention, mapping the conversion contribution representation value according to a preset mapping rule includes:

[0044] The above mapping transformation is performed according to the following expression:

[0045] Result_ T =Coeff*e Value

[0046] Among them, Result_ T represents the mapping transformation result, Coeff represents the preset coefficient, Coeff is greater than 1, and Value represents the conversion contribution representation value.

[0047] In one embodiment of the present invention, determining, from the candidate advertisements, a first advertisement whose advertisement features match the interest features of the first user includes:

[0048] determining a group of interested persons to which the first user belongs;

[0049] Obtaining a first advertiser of an advertisement of interest to a user identified by the user identity identifier in the determined interested group;

[0050] The advertisement commissioned by the first advertiser is selected from the candidate advertisements, and a first advertisement whose advertisement features match the interest features of the first user is obtained based on the selection result.

[0051] In one embodiment of the present invention, determining, from the candidate advertisements, a first advertisement whose advertisement features match the features of interest of the first user, includes:

[0052] Determining a second advertiser to which the candidate advertisement belongs;

[0053] Selecting a first advertiser with an advertising budget and an interested group from the second advertisers;

[0054] An advertisement among the candidate advertisements that belongs to the first advertiser and whose user identity identifier of the first user belongs to the interested group corresponding to the first advertiser is determined as the first advertisement.

[0055] In one embodiment of the present invention, after recalling the candidate advertisements, the method further includes:

[0056] Selecting a first number of candidate advertisements from the candidate advertisements in descending order of the push coefficients of the advertisements;

[0057] The step of determining, from the candidate advertisements, a first advertisement whose advertisement feature matches the feature of interest of the first user, comprises:

[0058] From the selected advertisements, determining a first advertisement whose advertisement characteristics match the characteristics of interest of the first user;

[0059] The step of ranking the candidate advertisements based on the adjusted push coefficients includes:

[0060] The selected advertisements are ranked based on the adjusted push coefficients.

[0061] In a second aspect, an embodiment of the present invention provides an advertisement ranking device, the device comprising:

[0062] An advertisement recall module, for recalling candidate advertisements in response to an advertisement exposure request;

[0063] An advertisement determination module, configured to determine, from the candidate advertisements, a first advertisement whose advertisement characteristics match the interest characteristics of a first user, wherein the first user is: a user identified by a user identity carried in the advertisement exposure request;

[0064] A coefficient obtaining module, used to obtain a push weighting coefficient of the first advertisement;

[0065] A coefficient adjustment module, configured to adjust the push coefficient of the first advertisement based on the push weighting coefficient;

[0066] The advertisement sorting module is used to sort the candidate advertisements based on the adjusted push coefficients.

[0067] In a third aspect, an embodiment of the present invention provides an electronic device, including a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other via the communication bus;

[0068] Memory, used to store computer programs;

[0069] The processor is used to implement any advertisement sorting method described in the first aspect when executing the program stored in the memory.

[0070] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the advertisement sorting method described in any one of the first aspects is implemented.

[0071] In a fifth aspect, an embodiment of the present invention provides a computer program product comprising instructions, which, when executed on a computer, enables the computer to execute any of the advertisement ranking methods described in the first aspect.

[0072] As can be seen from the above, when the scheme provided by the embodiment of the present invention is applied to sort advertisements, for the first advertisement, the push coefficient of the first advertisement is adjusted using the push weighting coefficient of the first advertisement, so that the push coefficient of the first advertisement can be increased, and the candidate advertisements are sorted based on the adjusted push coefficient, which can increase the probability of the first advertisement being pushed first. Since the advertising features of the first advertisement match the interest features of the first user, it means that the probability that the first user is interested in the first advertisement is high. In this case, once the first advertisement is pushed to the first user, the advertising conversion rate of the first advertisement can be improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0073] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art are briefly introduced below.

[0074] Figure 1 A flowchart of a method for determining a group of interested persons corresponding to an advertiser provided in an embodiment of the present invention;

[0075] Figure 2 A flowchart of a method for generating a push weighting coefficient provided by an embodiment of the present invention;

[0076] Figure 3 A flowchart of another method for generating a push weighting coefficient provided by an embodiment of the present invention;

[0077] Figure 4 A schematic diagram of a flow chart of an advertisement ranking method provided by an embodiment of the present invention;

[0078] Figure 5 A schematic diagram of a flow chart of another advertisement sorting method provided by an embodiment of the present invention;

[0079] Figure 6 A schematic diagram of the structure of an advertisement ranking device provided by an embodiment of the present invention;

[0080] Figure 7 A schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0081] The technical solutions in the embodiments of the present invention will be described below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0082] In order to increase the exposure of their products, various merchants will place advertisements on various advertising platforms to expand the audience of their products. Accordingly, various advertising promotion platforms will push advertisements to various users on behalf of advertisers. When pushing advertisements, the advertising promotion platform recalls the advertisements and pushes the advertisements to users based on the results of sorting the recalled advertisements. Therefore, the way advertisements are sorted directly affects what advertisements are pushed to users, and also directly affects the conversion rate of advertisements commissioned by advertisers. In view of this, the embodiments of the present invention provide an advertisement sorting method, device, electronic device and storage medium to improve the conversion rate of advertisements when pushing advertisements to users.

[0083] When applying the solution provided by the embodiment of the present invention to sort advertisements, after recalling the candidate advertisements, for the first advertisement whose advertising features among the candidate advertisements match the features of interest of the first user, the push coefficient of the first advertisement is adjusted using the push weighting coefficient of the first advertisement, and then the candidate advertisements are sorted based on the adjusted push coefficients.

[0084] On the basis of the above, the scheme provided in the embodiment of the present invention further establishes a connection between the push weighting coefficient and the advertiser. Specifically, the user identity identifiers of the users of the advertising promotion platform are divided into groups in advance, and the group of identified users who are interested in the advertisements of each advertiser are found from these groups. They are called the group of people interested in the advertisements of the advertiser, that is, the high-quality traffic group of the advertiser. Based on this, an association relationship is established between the group of people and the advertiser, and then based on the performance of the interested group in the advertisements commissioned by the advertiser, a push weighting coefficient is generated for the advertiser, that is, the push weighting coefficient of each advertisement commissioned by the advertiser is generated. In this way, when sorting the advertisements, the push coefficient of the advertisement commissioned by the advertiser can be increased based on the corresponding weighting coefficient of the advertiser, so that the sorting number of the advertisement can be moved forward, thereby increasing the chance of the advertisement being pushed to the user first.

[0085] On the basis of the above, after receiving the ad exposure request, the ad promotion platform adjusts the push coefficient corresponding to the ad based on the corresponding relationship between the above groups and advertisers and the push weight coefficient corresponding to the advertiser, and sorts the candidate ads. Of course, after the sorting is completed, the ad to be pushed can also be determined based on the sorting results, and the ad can be pushed to the advertiser's interested groups.

[0086] The above is only an overall description of the advertisement ranking scheme provided by the embodiment of the present invention, which will be described in detail below. In order to make the advertisement ranking scheme provided by the embodiment of the present invention easy to understand, some concepts involved in the scheme provided by the embodiment of the present invention and their acquisition methods are first described.

[0087] 1. Interested groups corresponding to advertisers

[0088] The above-mentioned interested groups refer to a set of user IDs of users who are interested in the advertisements commissioned by the advertiser. Each group includes multiple user IDs. A group can be an interested group corresponding to one advertiser or to multiple advertisers. In addition, one advertiser can also correspond to multiple interested groups.

[0089] The following is a description of a method for determining the interested groups corresponding to the advertiser.

[0090] In one embodiment of the present invention, see Figure 1 , a flow chart of a method for determining an interested group corresponding to an advertiser is provided, the method comprising the following steps S101-S104.

[0091] Step S101: classify the user identity identifier based on the historical behavior of the user identified by the user identity identifier to obtain a group of people corresponding to users with similar interests.

[0092] The objects that the users in the group of users with similar interests are interested in are similar. For example, the objects that the users are interested in belong to the same category, such as sports, food, etc. In one implementation, the objects that the users are interested in may be objects promoted by advertisements, so the group of users with similar interests is: the group of users with similar interests in advertisements.

[0093] In addition, the objects of interest to users may change with the changes in life needs and daily concerns. Therefore, the groups of people corresponding to the users with similar interests are not static. In order to ensure the accuracy of the groups used when pushing advertisements, in one embodiment of the present invention, the groups of people corresponding to the users with similar interests may be updated at fixed time intervals or at fixed times. For example, the groups may be updated every 48 hours or at 3 a.m. every day.

[0094] Furthermore, the historical behavior of the user identified by the user identity may be historical behavior related to advertising or historical behavior unrelated to advertising. Therefore, the population corresponding to the above-mentioned users with similar interests may be divided based on information directly related to advertising, such as users' browsing of pushed advertisements, or based on information indirectly related to advertising, such as other operations of users on the advertising promotion platform that are unrelated to advertising.

[0095] In one embodiment of the present invention, the user identity identifiers may be classified according to at least one of the following three characteristics of the historical behavior of the user identified by the user identity identifier to obtain a group of people corresponding to users with similar interests:

[0096] Feature 1: A first feature updated based on the user's operation behavior on the played advertisement in a recent historical period, wherein the first feature represents whether the user is interested in the advertisement.

[0097] The length of the above historical time period can be set according to specific needs, for example, 7 days, 10 days, etc.

[0098] The above-mentioned operation behavior may be clicking a skip button of the advertisement to skip the advertisement, maximizing the playback interface to play the advertisement, clicking to enter the purchase interface of the object promoted by the advertisement, etc.

[0099] Specifically, the number of times the above-mentioned various operation behaviors occur in a recent historical time period may be counted, and then the first feature is updated according to the number of times obtained by counting.

[0100] For example, a representation value can be set for each operation behavior, and then a weighted calculation is performed based on the representation value of the operation behavior and the number of times obtained by statistics to obtain an adjustment value of the first feature, and the above adjustment value is used to adjust the obtained first feature, thereby achieving an update of the first feature.

[0101] Since the user's operating behavior on the played advertisement is directly related to the user's feeling about the advertisement, it can directly reflect whether the user is interested in the advertisement. Therefore, based on the above-mentioned first feature, it is possible to more directly find users with similar interests in the advertisement, and then obtain the population corresponding to the users with similar interests in the advertisement, that is, the population corresponding to users with similar interests.

[0102] Feature 2: A second feature updated based on the user's operational behavior on the advertising promotion platform in a recent historical period, wherein the second feature represents the attributes of the user itself.

[0103] The above-mentioned advertising promotion platform may be a video platform, an audio platform, and the like.

[0104] When registering on an advertising promotion platform, users may fill in some of their own attributes, but for privacy reasons, some users may not be willing to fill in the detailed attributes. In addition, for an advertising promotion platform, its users are accounts, and the same account may be used by different people, for example, it may be used by family members, so the attributes of the user itself will also change. In view of the above situation, the above second feature can be continuously updated over time to obtain a more accurate second feature.

[0105] The user's operating behaviors on the advertising promotion platform may include browsing videos, playing audio, and member recharge, etc.

[0106] For example, at least one of the types, duration, and frequency of the above-mentioned operations performed by users on the advertising promotion platform in a recent historical period may be counted, and then the attributes of the user itself may be predicted based on the statistical results.

[0107] For example, you can count the types of videos that users have watched on a video platform in the past seven days and the length of time they have watched each video, and then predict the user's age, gender and other attributes based on the statistical types and lengths.

[0108] Since the interests and hobbies of users are often closely related to the attributes of the users themselves, the second feature mentioned above can indirectly reflect the interests and hobbies of users. Users with similar interests and hobbies are very likely to be interested in the same advertising objects. On this basis, considering the second feature when dividing the population corresponding to users with similar interests can enrich the information based on which the population is divided, thereby improving the accuracy of the division.

[0109] Feature 3: A third feature generated based on the interest prediction model, wherein the third feature represents the user's interest.

[0110] Specifically, the interest prediction model may be pre-trained to obtain features used in the interest prediction process using the interest prediction model as the third feature.

[0111] The user's interest represented by the third feature may be the type of advertisement object that the user is interested in, or may be the type of video, audio, news, etc. that the user is interested in.

[0112] Regardless of whether the interest represented by the third feature is directly or indirectly related to advertising, users with similar interests are very likely to be interested in the same advertising objects. When dividing the population corresponding to users with similar interests, introducing the third feature can increase the information based on which the population is divided, thereby improving the accuracy of the division.

[0113] After obtaining the above features, the similarity of the features between users can be calculated, and the identities of users corresponding to the features with similarity greater than a preset threshold can be classified into the same group of people.

[0114] At least one of the first feature, the second feature and the third feature is used to divide the population, so as to obtain a relatively prepared population corresponding to users with similar interests.

[0115] In addition, the above-mentioned method of dividing the population is based on the data recorded by the advertising promotion platform, that is, the population division is completed by mining user portraits based on the data recorded by the advertising promotion platform itself. In this way, a population with common characteristics can be constructed in combination with the business characteristics of the platform itself.

[0116] Step S102: statistics are collected on the interest representation values ​​of the groups in each group-advertiser pair towards the advertiser's advertisements during the sample time period to obtain corresponding statistical results for each group-advertiser pair.

[0117] A crowd-advertiser pair can be understood as an information pair formed by a crowd and an advertiser, and a crowd-advertiser pair includes a crowd and an advertiser. The crowd-advertiser pair is introduced in the embodiment of the present invention only for the convenience of description.

[0118] The above sample time period can be set according to needs, for example, it can be the last 7 days, the last month, etc.

[0119] For each user, he or she may express interest in the advertiser's advertisement in various dimensions. In view of this, in one embodiment of the present invention, the interest representation values ​​of the advertiser's advertisements expressed by the groups in each group-advertiser pair in the sample time period in the preset conversion evaluation dimension may be counted to obtain the corresponding statistical results of each group-advertiser pair in the preset conversion evaluation dimension. By performing statistics according to the preset conversion evaluation dimension, the statistical results can be made more targeted, thereby improving the targeting of the overall interest representation value obtained subsequently, and providing the accuracy of the determined advertiser's interested groups.

[0120] The above-mentioned preset conversion evaluation dimension may include one evaluation dimension or multiple evaluation dimensions. For specific descriptions, please refer to the subsequent embodiments, which will not be described in detail here.

[0121] During the advertising push process, the advertising promotion platform will record various information related to the advertising push. When performing the above statistics, according to the meaning of the conversion evaluation dimension, the information related to the user identified by each user identity in the crowd and the conversion evaluation dimension can be obtained from the information recorded by the advertising promotion platform, and then statistics can be performed.

[0122] Step S103: based on the corresponding statistical results of each group-advertiser pair, the overall interest representation value of the group in each group-advertiser pair towards the advertiser's advertisement is obtained.

[0123] After statistics, corresponding statistical results will be obtained for each group-advertiser pair, for example, the statistical results corresponding to the group-advertiser pair under various dimensions. In view of this, when statistics are performed under the preset conversion evaluation dimension, the corresponding statistical results under each conversion evaluation dimension will be obtained for each group-advertiser pair. On this basis, the overall interest representation value of the group in the advertiser's advertisement in the group-advertiser pair can be obtained in different ways. For example:

[0124] In one implementation, when multiple conversion evaluation dimensions are included, weighted calculation may be performed on the corresponding statistical results of the crowd advertisers under each conversion evaluation dimension, and then the calculation result is determined as the above-mentioned overall interest representation value.

[0125] In another implementation, when multiple conversion evaluation dimensions are included, the maximum statistical result, the median of the statistical results, etc. corresponding to the group advertisers under each conversion evaluation dimension can be determined as the above-mentioned overall interest representation value.

[0126] Step S104: Determine that the group of people in the advertiser's group whose overall interest representation value is greater than a preset interest threshold is the advertiser's interested group.

[0127] From the above, it can be seen that in the solution provided by this embodiment, based on the interest representation value of the crowd in the advertiser's advertisement, it is determined whether the crowd with similar interests is the interested crowd of each advertiser. In this way, each advertiser can find the crowd that is interested in its advertisement, thereby improving the accuracy of crowd search.

[0128] The preset conversion evaluation dimensions are explained below.

[0129] In one embodiment of the present invention, the preset conversion evaluation dimension includes at least one of the following dimensions 1 to 4:

[0130] Dimension 1: User interest in target categories, where target categories are the categories to which the advertiser's advertisements belong.

[0131] For example, if the advertiser's advertising target is game X, then the target category is game; if the advertiser's advertising target is shampoo, then the target category is daily necessities.

[0132] Under this dimension, the values ​​representing the interest of users in the population in the advertisements in the target category can be the number of clicks, ECPM (Effective Cost Per Mile, the advertising revenue that can be obtained for every thousand impressions), whether the user pays, the payment value, the payment efficiency, etc.

[0133] Dimension 2: Parameters that directly describe whether users convert the advertiser's ads.

[0134] In this dimension, the above parameters may be the number of clicks, ECPM, etc., and the interest representation value of the user identified by the user identity in the corresponding population in the advertiser's advertisement performance may be the number of clicks, the ECPM, etc.

[0135] Dimension 3: Parameters that indirectly describe whether users convert the advertiser's ads.

[0136] Under this dimension, the above parameters can be whether the user has paid, the payment value, the payment efficiency, etc. The interest representation value of the user identified by the user identity in the corresponding population in the advertiser's advertising performance can be the identification value of whether the user has paid, the specific value of the payment value, the efficiency value of the payment efficiency, etc.

[0137] Dimension 4: The sales revenue brought to the advertiser by the user who is the target of the advertiser’s advertising.

[0138] As can be seen from the above, the conversion evaluation dimension in the solution provided in this embodiment can include multiple dimensions. In this way, when looking for interested groups for advertisers, it is possible to measure from multiple angles whether users in the group are interested in the advertiser's advertisements, thereby improving the accuracy of finding interested groups for advertisers.

[0139] 2. Push weighting coefficient

[0140] After recalling the advertisements, the advertising promotion platform needs to sort the recalled advertisements according to the push coefficient, and then select the advertisements to be pushed from the sorted advertisements, and then push them. The above push weighting coefficient is used to adjust the push coefficient and thus increase the push coefficient.

[0141] Among them, the above-mentioned push coefficient can be the bid of the advertisement, so the above-mentioned push weighted coefficient can be understood as the price increase coefficient.

[0142] Of course, the push coefficient may also be other values, such as the push priority agreed upon by the advertiser and the advertising promotion platform, etc., and the embodiment of the present invention is not limited to this.

[0143] Each advertisement has a corresponding push coefficient, and each corresponding advertisement also has a corresponding push weighting coefficient. In one implementation, the push weighting coefficients of the advertisements of the same advertiser can be the same, so when generating the push weighting coefficient, it can be generated on an advertiser basis, that is, a push weighting coefficient is generated for each advertiser. After generating a push weighting coefficient for an advertiser, the weighting coefficient can be stored in a database, so that when the advertisement is pushed subsequently, it can be directly read from the database, which can greatly improve the efficiency of advertisement push. Among them, the above-mentioned database can store data based on the Key-Value mode.

[0144] In addition, the push weighting coefficient corresponding to each advertiser can be updated according to a set period.

[0145] In one embodiment of the present invention, the push weighting coefficient can be determined based on the interest characterization value corresponding to the first interested group, wherein the first interested group is: a set of user identity identifiers of users who are interested in the advertisement commissioned by the first advertiser, and the interest characterization value represents the degree of interest of the users identified by the user identity identifiers in the first interested group in the advertisement commissioned by the first advertiser, and the first advertiser is: the advertiser of the first advertisement. In this way, the push weighting coefficient can be associated with the advertisement commissioned by the advertiser, and then when adjusting the push coefficient of the advertisement, it can be adjusted in a targeted manner based on whether the user is interested in the advertisement commissioned by the advertiser, so that the adjusted push coefficient is more accurate, and thus the accuracy of ranking the candidate advertisements can be improved.

[0146] On the basis of the above, further, in one embodiment of the present invention, the above push weighting coefficient is determined based on the interest representation value corresponding to the first interested group under the preset conversion evaluation dimension. In this way, the push weighting coefficient is obtained based on the data under the set conversion evaluation dimension, so that the data used in determining the push weighting coefficient is more targeted, thereby further improving the accuracy of the determined push weighting coefficient.

[0147] Furthermore, the method for generating the corresponding weighting coefficient for each advertiser is the same. The method for generating the push weighting coefficient is described below by taking the generation of the weighting coefficient corresponding to the first advertiser as an example.

[0148] In one embodiment of the present invention, see Figure 2 , a flow chart of a method for generating a push weighting coefficient is provided, the method comprising the following steps S201-S204.

[0149] Step S201: obtaining the number of advertisement conversions brought to a first advertiser by a user identified by a user identity in a first interested group in a first time period.

[0150] The duration of the first time period may be set according to demand, for example, it may be 7 days, 15 days, etc.

[0151] As mentioned above, during the process of advertising push, the advertising promotion platform will record the information generated by the advertising push. Based on this, the above-mentioned advertising conversion number can be obtained from the information recorded by the platform.

[0152] Step S202: obtaining interest representation values ​​expressed by users identified by user identities in a first interested group in an advertisement commissioned by a first advertiser within a first time period.

[0153] Similar to the above steps, in this step, the interest representation value expressed by the user identified by the user identity identifier in the first interested group in the advertisement commissioned by the first advertiser during the first time period can also be obtained from the data recorded by the platform.

[0154] Specifically, the interest representation value of the user identified by the user identity identifier in the first interested group in the advertisement commissioned by the first advertiser within the first time period under the preset conversion evaluation dimension may be obtained.

[0155] Step S203: Based on the obtained interest representation value, the above-mentioned advertisement conversion number and the expected value of the preset conversion index, determine the conversion contribution representation value of the user identified by the user identity in the first interested group to the first advertiser.

[0156] The above preset conversion indicator may be RPM (Revenue Per Mille, revenue obtained per thousand ad impressions).

[0157] The above conversion contribution representation value can be understood as: the contribution brought by the user identified by the user identity identifier in the first interested group to the conversion of the advertisement of the first advertiser.

[0158] In one embodiment of the present invention, the obtained interest representation value and the expected value of the preset conversion index can be normalized first; if the above-mentioned advertising conversion number is greater than the preset quantity threshold, an adjustment value is generated based on the preset adjustment coefficient and the normalized expected value, and the normalized interest representation value is adjusted based on the adjustment value to obtain the conversion contribution representation value of the user identified by the user identity in the first interested group to the first advertiser; if the above-mentioned advertising conversion number is not greater than the preset quantity threshold, an adjustment value is generated based on the preset adjustment coefficient and the normalized expected value as the conversion contribution representation value. After the above-mentioned normalization process is performed, even if the data difference between the obtained interest representation value and the expected value is large, it will not affect the final conversion contribution representation value. In addition, determining the conversion contribution representation value in different ways can improve the accuracy of the obtained conversion contribution representation value.

[0159] Specifically, the obtained interest representation value may be normalized in the following manner:

[0160] The interest representation value of the user identified by the user identity in the first interested group in the first time period for the advertisement in the target category under the preset conversion evaluation dimension is obtained, recorded as the first representation value, and then the interest representation value obtained above is normalized according to the following expression:

[0161] Obtained interest characterization value / first characterization value - 1

[0162] Similar to the above, the expected value of the preset conversion indicator can be normalized as follows:

[0163] Obtain an index value of the user identified by the user identity in the first interested group under the preset conversion index for the advertisement in the target category within the first time period, recorded as the first index value, and then normalize the above expected value according to the following expression:

[0164] The above expected value / the first indicator value - 1

[0165] In one implementation, the adjustment value may be equal to the product of a preset adjustment coefficient and a normalized expected value. Correspondingly, when an adjustment value is generated based on the preset adjustment coefficient and the normalized expected value and used as a conversion contribution representation value, the adjustment value may be equal to the product of the preset adjustment coefficient and the normalized expected value as the conversion contribution representation value.

[0166] In another implementation, adjusting the normalized interest representation value based on the adjustment value may be performed by adding the adjustment value to the normalized interest representation value.

[0167] Step S204: Generate a push weighting coefficient based on the above conversion contribution representation value.

[0168] The larger the above-mentioned conversion contribution representation value is, the greater the conversion contribution of the user identified by the user identity in the first interested group to the first advertiser is, so the push weighting coefficient can be increased. Conversely, the smaller the above-mentioned conversion contribution representation value is, the smaller the conversion contribution of the user identified by the user identity in the first interested group to the first advertiser is, so the push weighting coefficient can be decreased.

[0169] In one embodiment of the present invention, the push weighting coefficient may be generated in the following manner:

[0170] If the conversion contribution representation value is greater than the preset contribution threshold, the conversion contribution representation value is mapped and transformed according to the preset mapping rule, and the minimum value between the mapping transformation result and the preset maximum push weighting coefficient is determined as the push weighting coefficient; if the conversion contribution representation value is not greater than the preset contribution threshold and is greater than 0, the push weighting coefficient is determined to be the first value; if the conversion contribution representation value is less than or equal to 0, the push weighting coefficient is determined to be the second value, wherein the second value is less than the first value. In this way, the push weighting coefficient can be determined in a targeted manner according to the different contributions of the first interested group to the advertisement conversion of the first advertiser, which can improve the accuracy of the determined push weighting coefficient.

[0171] Specifically, the above mapping transformation can be performed according to the following expression:

[0172] Result_ T =Coeff*e Value

[0173] Among them, Result_ T Indicates the mapping transformation result, Coeff indicates the preset coefficient, Coeff is greater than 1, and Value indicates the conversion contribution representation value.

[0174] The first value and the second value may be set according to specific requirements. For example, the first value may be 1 and the second value may be 0.

[0175] From the above, it can be seen that in the solution provided by this embodiment, when determining the push weighting coefficient of the first advertisement, the conversion contribution of the user identified by the user identity identifier in the interested group of the advertisement entrusted by the first advertiser to the first advertiser is taken into consideration. In this way, the push weighting coefficient can be increased when the conversion contribution is large, thereby increasing the probability that the advertisement of the first advertiser will be pushed first in the advertisement sorting process. Since the user identified by the user identity identifier in the interested group of the advertisement entrusted by the first advertiser has a high conversion contribution to the first advertiser, the probability of the advertisement of the first advertiser being converted is further increased.

[0176] In addition to generating the push weighting coefficient in the above manner, it is also possible to consider whether the push weighting coefficient corresponding to the advertisement entrusted by the first advertiser has been enabled for advertisement push, the number of days of activation, etc. Based on this, in another embodiment of the present invention, see Figure 3 , a flow chart of another method for generating a push weighting coefficient is provided, the method comprising the following steps S301-S309.

[0177] Step S301: Obtain the number of advertisement conversions brought to a first advertiser by a user identified by a user identity in a first interested group in a first time period.

[0178] Step S302: obtaining the interest representation value of the user identified by the user identity identifier in the first interested group in the advertisement commissioned by the first advertiser within the first time period under the preset conversion evaluation dimension.

[0179] Step S303: Based on the obtained interest representation value, the number of advertisement conversions and the expected value of the preset conversion indicator, determine the conversion contribution representation value of the user identified by the user identity in the first interested group to the first advertiser.

[0180] Step S304: Determine whether the push weighting coefficient corresponding to the first advertiser has been enabled for advertising push. If not enabled and the conversion contribution representation value is greater than the closing threshold, execute the following step S305; if enabled, execute the following step S306.

[0181] Step S305: mapping and transforming the conversion contribution representation value according to a preset mapping rule, and determining the minimum value between the mapping and transformation result and the preset maximum push weighting coefficient as the push weighting coefficient.

[0182] Step S306: Determine whether the number of days for enabling the promotion weighting coefficient corresponding to the first advertiser to push advertisements meets the duration of the first time. If not, execute the above step S305. If yes, execute the following S307.

[0183] Step S307: Determine whether the number of advertising conversions is greater than a preset threshold value. If not, execute S308. If greater than the preset threshold value and the conversion contribution representation value is greater than the preset contribution threshold, execute S305. If greater than the preset threshold value and the conversion contribution representation value is not greater than the preset contribution threshold and is greater than 0, execute S309.

[0184] Step S308: Determine that the push weighting coefficient is a second value.

[0185] Step S309: Determine that the push weighting coefficient is a first value.

[0186] Based on the above description, the advertisement ranking method provided by the embodiment of the present invention is described in detail below.

[0187] See also Figure 4 , a flow chart of an advertisement ranking method is provided, the method comprising the following steps S401-S405.

[0188] Step S401: Recall candidate advertisements in response to an advertisement exposure request.

[0189] The above-mentioned advertising exposure request is used to request that an advertisement be pushed to a user. It may be triggered by a user's operation on the platform, for example, a user requests to play a video, or it may be triggered by reaching a preset advertising push timing, for example, the same video has been played continuously for 20 minutes.

[0190] Step S402: Determine, from the selected advertisements, a first advertisement whose advertisement characteristics match the characteristics of interest of the first user.

[0191] The first user is: a user identified by a user identity carried in the advertisement exposure request.

[0192] In one implementation, the interested group to which the first user belongs can be determined, the first advertiser of the advertisement to which the user identified by the user identity in the determined interested group belongs is obtained, the advertisement commissioned by the first advertiser is selected from the candidate advertisements, and the first advertisement whose advertisement features match the interest features of the first user is obtained based on the selection result. In this way, the first advertisement that the first user is sensitive to can be accurately determined from the candidate advertisements, thereby improving the accuracy of the determined first advertisement.

[0193] Specifically, each interested group may be set with a group identifier. When a user's user identity identifier belongs to a certain group, the group identifier of the group may be added to the user's attributes. In this way, the interested group to which the first user belongs may be determined based on the group identifier recorded in the first user's attributes.

[0194] In addition, the population identification of each advertiser's interested population can be recorded, thus establishing a correspondence between advertisers and populations. On this basis, the first advertiser whose advertisement is of interest to the user identified by the user identification in the determined interested population can be obtained based on the correspondence between the advertisers and populations.

[0195] Furthermore, the first advertiser may entrust multiple advertisements to the advertising promotion platform, and these advertisements may belong to multiple types, and the types of advertisements that each user is interested in may be different. In view of this, in one implementation, after selecting the advertisement entrusted by the first advertiser from the candidate advertisements, when obtaining the first advertisement whose advertisement features match the interest features of the first user based on the selection results, the advertisement whose advertisement type is the type that the first user is interested in can be searched from the selection results as the first advertisement. Of course, it is also possible to ignore the advertisement type and directly use all advertisements in the selection results as the first advertisement. The embodiments of the present invention are not limited to this.

[0196] In another implementation, the second advertiser to which the candidate advertisement belongs may be determined, and the first advertiser with an advertisement budget and an interested group may be selected from the second advertisers, and the advertisement belonging to the first advertiser and the user identity of the first user belonging to the interested group corresponding to the first advertiser may be determined as the first advertisement. In this way, the subsequent push coefficient adjustment operation is performed only for the advertisers with an advertisement budget, rather than for all advertisers, thereby achieving targeted adjustment.

[0197] Step S403: Obtain the push weighting coefficient of the first advertisement.

[0198] In one implementation, the push weighting coefficient may be determined based on the interest representation value corresponding to the first interested group. Further, the push weighting coefficient is determined based on the interest representation value corresponding to the first interested group under a preset conversion evaluation dimension.

[0199] The method for generating the push weighting coefficient can refer to the above-mentioned embodiment and will not be repeated here.

[0200] Specifically, the push weighting coefficient of the first advertiser may be obtained from a database storing the push weighting coefficients of various advertisers as the push weighting coefficient of the first advertisement.

[0201] Step S404: Based on the push weighting coefficient, adjust the push coefficient of the first advertisement.

[0202] When adjusting the push coefficient of the first advertisement, the adjustment may be performed by multiplying the push weighting coefficient by the push coefficient of the first advertisement.

[0203] Step S405: sorting the candidate advertisements based on the adjusted push coefficients.

[0204] When selecting an advertisement to be pushed from candidate advertisements, the advertisements may be sorted in descending order according to the push coefficients.

[0205] As can be seen from the above, when the scheme provided by the embodiment of the present invention is applied to sort advertisements, for the first advertisement, the push coefficient of the first advertisement is adjusted using the push weighting coefficient of the first advertisement, so that the push coefficient of the first advertisement can be increased, and the candidate advertisements are sorted based on the adjusted push coefficient, which can increase the probability of the first advertisement being pushed first. Since the advertising features of the first advertisement match the interest features of the first user, it means that the probability that the first user is interested in the first advertisement is high. In this case, once the first advertisement is pushed to the first user, the advertising conversion rate of the first advertisement can be improved.

[0206] In another embodiment of the present invention, see Figure 5, provides a flow chart of another advertisement ranking method. In this embodiment, relative to Figure 4 The illustrated embodiment, after recalling the candidate advertisements in step S401, further includes:

[0207] Step S406: Select the first number of candidate advertisements from the candidate advertisements in descending order of the push coefficients of the advertisements.

[0208] In this case, the above step S402 includes:

[0209] Step S402A: Determine, from the selected advertisements, a first advertisement whose advertisement characteristics match the characteristics of interest of the first user.

[0210] The above step S405 includes:

[0211] Step S405A: Sort the selected advertisements based on the adjusted push coefficients.

[0212] In this way, after the advertisement is recalled, the advertisements are sorted and screened for the first time. Then, the push coefficient of the first advertisement of the first advertiser in the screening results is adjusted, and then a secondary screening is performed. This allows advertisements that already have a higher push priority to have an even higher priority and be pushed first.

[0213] Corresponding to the above-mentioned advertisement ranking method, an embodiment of the present invention further provides an advertisement ranking device.

[0214] See also Figure 6 , a schematic diagram of the structure of an advertisement sorting device is provided, the device comprising:

[0215] Advertisement recall module 601, used to recall candidate advertisements in response to advertisement exposure requests;

[0216] The advertisement determination module 602 is used to determine, from the candidate advertisements, a first advertisement whose advertisement characteristics match the interest characteristics of the first user, wherein the first user is: a user identified by a user identity carried in the advertisement exposure request;

[0217] The coefficient obtaining module 603 is used to obtain the push weighting coefficient of the first advertisement;

[0218] A coefficient adjustment module 604, configured to adjust the push coefficient of the first advertisement based on the push weighting coefficient;

[0219] The advertisement ranking module 605 is used to rank the candidate advertisements based on the adjusted push coefficients.

[0220] As can be seen from the above, when the scheme provided by the embodiment of the present invention is applied to sort advertisements, for the first advertisement, the push coefficient of the first advertisement is adjusted using the push weighting coefficient of the first advertisement, so that the push coefficient of the first advertisement can be increased, and the candidate advertisements are sorted based on the adjusted push coefficient, which can increase the probability of the first advertisement being pushed first. Since the advertising features of the first advertisement match the interest features of the first user, it means that the probability that the first user is interested in the first advertisement is high. In this case, once the first advertisement is pushed to the first user, the advertising conversion rate of the first advertisement can be improved.

[0221] In one embodiment of the present invention, the push weighting coefficient is determined based on the interest representation value corresponding to the first interested group, the first interested group is: a set of user identity identifiers of users who are interested in the advertisement commissioned by the first advertiser, the interest representation value represents the degree of interest of the users identified by the user identity identifiers in the first interested group in the advertisement commissioned by the first advertiser, and the first advertiser is: the advertiser of the first advertisement.

[0222] In one embodiment of the present invention, the advertiser's interested groups are determined through the following modules:

[0223] A similar group determination module is used to classify the user identity identifiers based on the historical behaviors of the users identified by the user identity identifiers, and obtain groups corresponding to users with similar interests;

[0224] The representation value statistics module is used to count the interest representation values ​​of the advertiser's advertisement performance of the population in each population-advertiser pair within the sample time period, and obtain the corresponding statistical results of each population-advertiser pair;

[0225] An overall representation value obtaining module, for obtaining an overall interest representation value of the group of people in each group of advertisers for the advertisement of the advertiser based on the corresponding statistical results of each group of advertisers;

[0226] The module for determining the group of people of interest is used to determine that the group of people in the advertiser's group of people whose overall interest representation value is greater than a preset interest threshold is the group of people of interest to the advertiser.

[0227] In one embodiment of the present invention, the push weighting coefficient is determined based on the interest representation value corresponding to the first interested group under the preset conversion evaluation dimension;

[0228] The characterization value statistics module is specifically used to count the interest characterization values ​​of each group of advertisers in the advertiser's advertisements under the preset conversion evaluation dimension during the sample time period, and obtain the corresponding statistical results of each group of advertisers under the preset conversion evaluation dimension.

[0229] In one embodiment of the present invention, the conversion evaluation dimension includes at least one of the following dimensions:

[0230] The user's interest in the target category, where the target category is: the category to which the advertiser's advertisement belongs;

[0231] Parameters that directly describe whether a user converts an advertiser's ad;

[0232] Parameters that indirectly describe whether a user converts an advertiser's ad;

[0233] The sales revenue brought to advertisers by the targeted users.

[0234] In one embodiment of the present invention, the similar group determination module is specifically used to classify the user identity identifier according to at least one of the following characteristics of the historical behavior of the user identified by the user identity identifier, and obtain the group corresponding to the user with similar interests:

[0235] A first feature updated according to the user's operation behavior on the played advertisement in a recent historical period, wherein the first feature represents whether the user is interested in the advertisement;

[0236] A second feature updated according to the user's operation behavior on the advertising promotion platform in a recent historical period, wherein the second feature represents an attribute of the user itself;

[0237] A third feature is generated based on the interest prediction model, wherein the third feature represents the interest of the user.

[0238] In one embodiment of the present invention, the push weighting coefficient is generated by the following modules:

[0239] A conversion number obtaining module, used to obtain the number of advertisement conversions brought to the first advertiser by the user identified by the user identity identifier in the first interested group in the first time period;

[0240] An interest representation value obtaining module, used to obtain an interest representation value expressed by a user identified by a user identity identifier in a first interested group for an advertisement commissioned by a first advertiser within a first time period;

[0241] A conversion contribution representation value determination module, configured to determine a conversion contribution representation value of a user identified by a user identity in a first interested group to a first advertiser based on the obtained interest representation value, the number of advertisement conversions, and an expected value of a preset conversion indicator;

[0242] The coefficient generation module is used to generate a push weighting coefficient based on the conversion contribution representation value.

[0243] In one embodiment of the present invention, the conversion contribution representation value determination module is specifically used to normalize the obtained interest representation value and the expected value of the preset conversion indicator; if the number of advertising conversions is greater than the preset quantity threshold, an adjustment value is generated based on the preset adjustment coefficient and the normalized expected value, and the normalized interest representation value is adjusted based on the adjustment value to obtain the conversion contribution representation value of the user identified by the user identity in the first interested group to the first advertiser; if the number of advertising conversions is not greater than the preset quantity threshold, an adjustment value is generated based on the preset adjustment coefficient and the normalized expected value as the conversion contribution representation value.

[0244] In one embodiment of the present invention, the coefficient generation module is specifically used to map the conversion contribution representation value according to a preset mapping rule if the conversion contribution representation value is greater than a preset contribution threshold, and determine the minimum value between the mapping transformation result and the preset maximum push weighting coefficient as the push weighting coefficient; if the conversion contribution representation value is not greater than the preset contribution threshold and is greater than 0, determine the push weighting coefficient to be a first value; if the conversion contribution representation value is less than or equal to 0, determine the push weighting coefficient to be a second value, wherein the second value is less than the first value.

[0245] In one embodiment of the present invention, the coefficient generation module is specifically used to perform the above mapping transformation according to the following expression:

[0246] Result_ T =Coeff*e Value

[0247] Among them, Result_ T represents the mapping transformation result, Coeff represents the preset coefficient, Coeff is greater than 1, and Value represents the conversion contribution representation value.

[0248] In one embodiment of the present invention, the advertisement determination module 602 is specifically used to determine the interested group to which the first user belongs; obtain the first advertiser whose advertisement is of interest to the user identified by the user identity in the determined interested group; select the advertisement commissioned by the first advertiser from the candidate advertisements, and obtain the first advertisement whose advertisement characteristics match the interest characteristics of the first user based on the selection result.

[0249] In one embodiment of the present invention, the above-mentioned advertisement determination module 602 is specifically used to determine the second advertiser to which the candidate advertisement belongs; select a first advertiser with an advertising budget and an interested group from the second advertisers; and determine, as the first advertisement, the advertisement among the candidate advertisements that belongs to the first advertiser and in which the user identity identifier of the first user belongs to the interested group corresponding to the first advertiser.

[0250] In one embodiment of the present invention, the advertisement ranking device further includes:

[0251] The advertisement selection module is used to select the first number of candidate advertisements from the candidate advertisements in descending order of the push coefficients of the advertisements after the candidate advertisements are recalled by the advertisement recall module;

[0252] The advertisement determination module 602 is specifically configured to determine a first advertisement that the first user is interested in from among the selected advertisements;

[0253] The advertisement ranking module 605 is specifically configured to rank the selected advertisements based on the adjusted push coefficients.

[0254] The beneficial effects corresponding to the above-mentioned embodiments can be found in the above-mentioned method embodiment part and will not be described in detail here.

[0255] Corresponding to the above-mentioned advertisement ranking method, an embodiment of the present invention further provides an electronic device, a computer-readable storage medium and a computer program product.

[0256] See also Figure 7 , provides a structural diagram of an electronic device, including a processor 701, a communication interface 702, a memory 703 and a communication bus 704, wherein the processor 701, the communication interface 702, and the memory 703 communicate with each other through the communication bus 704,

[0257] Memory 703, used for storing computer programs;

[0258] The processor 701 is used to implement the advertisement sorting scheme provided by the above method embodiment when executing the program stored in the memory 703.

[0259] The communication bus mentioned in the above terminal can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. The communication bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, only one thick line is used in the figure, but it does not mean that there is only one bus or one type of bus.

[0260] The communication interface is used for communication between the above terminal and other devices.

[0261] The memory may include a random access memory (RAM) or a non-volatile memory, such as at least one disk memory. Optionally, the memory may also be at least one storage device located away from the aforementioned processor.

[0262] The above-mentioned processor can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.

[0263] In another embodiment of the present invention, a computer-readable storage medium is provided, wherein a computer program is stored in the computer-readable storage medium, and when the computer program is executed by a processor, the advertisement ranking scheme provided by the above method embodiment is implemented.

[0264] In another embodiment provided by the present invention, a computer program product including instructions is also provided, which, when executed on a computer, enables the computer to execute the advertisement ranking scheme provided by the above method embodiment.

[0265] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented by software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in the embodiment of the present invention is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from a website site, computer, server or data center to another website site, computer, server or data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more available media integrated. The available medium can be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid-state hard disk Solid State Disk (SSD)), etc.

[0266] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0267] Each embodiment in this specification is described in a related manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device, computer-readable storage medium, and computer program product embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiments.

[0268] The above description is only a preferred embodiment of the present invention and is not intended to limit the protection scope of the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention are included in the protection scope of the present invention.

Claims

1. An advertisement ranking method, characterized in that: The method comprises: Recalling candidate advertisements in response to an advertisement exposure request; Determine, from the candidate advertisements, a first advertisement whose advertisement feature matches a feature of interest of a first user, wherein the first user is: a user identified by a user identity carried in the advertisement exposure request; Obtaining a push weighting coefficient for the first advertisement; Based on the push weighting coefficient, adjusting the push coefficient of the first advertisement; The candidate advertisements are ranked based on the adjusted push coefficients.

2. The method according to claim 1, characterized in that The push weighting coefficient is determined based on the interest representation value corresponding to the first interested group, the first interested group is: a set of user identity identifiers of users who are interested in the advertisement commissioned by the first advertiser, the interest representation value represents the degree of interest of the users identified by the user identity identifiers in the first interested group in the advertisement commissioned by the first advertiser, and the first advertiser is: the advertiser of the first advertisement.

3. The method according to claim 2, characterized in that The interested groups corresponding to the advertiser are determined in the following way: Based on the historical behavior of the user identified by the user identity identifier, the user identity identifier is classified to obtain a group of people corresponding to users with similar interests; The interest representation values ​​of the groups in each group-advertiser pair towards the advertiser's advertisements within the sample time period are counted to obtain the corresponding statistical results of each group-advertiser pair; Based on the corresponding statistical results of each group-advertiser pair, the overall interest representation value of the group in each group-advertiser pair for the advertiser's advertisement is obtained; It is determined that the population in the advertiser pair whose overall interest representation value is greater than a preset interest threshold is the advertiser's interested population.

4. The method according to claim 3, characterized in that The push weighting coefficient is determined based on the interest representation value corresponding to the first interested group under the preset conversion evaluation dimension; The statistical analysis of the interest representation values ​​of the groups of people in each group-advertiser pair in the sample time period to the advertiser's advertisements is performed to obtain the corresponding statistical results of each group-advertiser pair, including: Statistics are collected on the interest representation values ​​of the advertisers in each group of advertisers in the preset conversion evaluation dimension during the sample period to obtain the corresponding statistical results of the advertisers in each group of advertisers in the preset conversion evaluation dimension.

5. The method according to claim 4, characterized in that The preset conversion evaluation dimension includes at least one of the following dimensions: The user's interest in the target category, wherein the target category is: the category to which the advertiser's advertisement belongs; Parameters that directly describe whether a user converts an advertiser's ad; Parameters that indirectly describe whether a user converts an advertiser's ad; The sales revenue brought to the advertiser by the target user.

6. The method according to claim 3, characterized in that The user identity identifiers are classified based on the historical behaviors of the users identified by the user identity identifiers to obtain the groups of users corresponding to users with similar interests, including: The user identity is classified according to at least one of the following characteristics of the historical behavior of the user identified by the user identity to obtain a group of users with similar interests: A first feature updated according to the user's operation behavior on the played advertisement in a recent historical period, wherein the first feature represents whether the user is interested in the advertisement; A second feature updated according to the operation behavior of the user on the advertising promotion platform in the recent historical time period, wherein the second feature represents an attribute of the user itself; A third feature is generated based on the interest prediction model, wherein the third feature represents the interest of the user.

7. The method according to claim 2, characterized in that The push weighting coefficient is generated in the following way: Obtaining the number of advertisement conversions brought to the first advertiser by the user identified by the user identity identifier in the first interested group in the first time period; Obtaining an interest representation value expressed by a user identified by a user identity identifier in the first interested group in the advertisement commissioned by the first advertiser during the first time period; Determine, based on the obtained interest representation value, the number of advertisement conversions, and an expected value of a preset conversion indicator, a conversion contribution representation value of the user identified by the user identity in the first interested group to the first advertiser; The push weighting coefficient is generated based on the conversion contribution representation value.

8. The method according to claim 7, characterized in that The determining, based on the obtained interest representation value, the number of advertisement conversions, and the expected value of a preset conversion indicator, a conversion contribution representation value of the user identified by the user identity in the first interested group to the first advertiser includes: Normalizing the obtained interest representation value and the expected value of the preset conversion indicator; If the number of advertisement conversions is greater than a preset number threshold, an adjustment value is generated based on a preset adjustment coefficient and a normalized expected value, and the normalized interest representation value is adjusted based on the adjustment value to obtain a conversion contribution representation value of the user identified by the user identity identifier in the first interested group to the first advertiser; If the number of advertisement conversions is not greater than the preset threshold value, an adjustment value is generated based on a preset adjustment coefficient and a normalized expected value as a conversion contribution representation value.

9. The method according to claim 7, characterized in that: The generating the push weighting coefficient based on the conversion contribution representation value includes: If the conversion contribution representation value is greater than a preset contribution threshold, mapping transformation is performed on the conversion contribution representation value according to a preset mapping rule, and the minimum value between the mapping transformation result and a preset maximum push weighting coefficient is determined as the push weighting coefficient; If the conversion contribution representation value is not greater than the preset contribution threshold and is greater than 0, determining the push weighting coefficient to be a first value; If the conversion contribution representation value is less than or equal to 0, the push weighting coefficient is determined to be a second value, wherein the second value is less than the first value.

10. The method according to claim 9, characterized in that The mapping transformation of the conversion contribution representation value according to a preset mapping rule includes: The above mapping transformation is performed according to the following expression: Result_ T =Coeff*e Value Among them, Result_ T represents the mapping transformation result, Coeff represents the preset coefficient, Coeff is greater than 1, and Value represents the conversion contribution representation value.

11. The method according to claim 2, characterized in that The step of determining, from the candidate advertisements, a first advertisement whose advertisement characteristics match the interest characteristics of the first user, comprises: determining a group of interested persons to which the first user belongs; Obtaining a first advertiser of an advertisement of interest to a user identified by the user identity identifier in the determined interested group; The advertisement commissioned by the first advertiser is selected from the candidate advertisements, and a first advertisement whose advertisement features match the interest features of the first user is obtained based on the selection result.

12. The method according to any one of claims 1 to 11, characterized in that The step of determining, from the candidate advertisements, a first advertisement whose advertisement feature matches the feature of interest of the first user, comprises: Determining a second advertiser to which the candidate advertisement belongs; Selecting a first advertiser with an advertising budget and an interested group from the second advertisers; An advertisement among the candidate advertisements that belongs to the first advertiser and whose user identity identifier of the first user belongs to the interested group corresponding to the first advertiser is determined as the first advertisement.

13. The method according to any one of claims 1 to 11, characterized in that Following the recall candidate advertisement, there is also included: Selecting a first number of candidate advertisements from the candidate advertisements in descending order of the push coefficients of the advertisements; The step of determining, from the candidate advertisements, a first advertisement whose advertisement feature matches the feature of interest of the first user, comprises: From the selected advertisements, determining a first advertisement whose advertisement characteristics match the characteristics of interest of the first user; The step of ranking the candidate advertisements based on the adjusted push coefficients includes: The selected advertisements are ranked based on the adjusted push coefficients.

14. An advertisement ranking device, characterized in that: The device comprises: An advertisement recall module, for recalling candidate advertisements in response to an advertisement exposure request; An advertisement determination module, configured to determine, from the candidate advertisements, a first advertisement whose advertisement characteristics match the interest characteristics of a first user, wherein the first user is: a user identified by a user identity carried in the advertisement exposure request; A coefficient obtaining module, used to obtain a push weighting coefficient of the first advertisement; A coefficient adjustment module, configured to adjust the push coefficient of the first advertisement based on the push weighting coefficient; The advertisement sorting module is used to sort the candidate advertisements based on the adjusted push coefficients.

15. An electronic device, characterized in that: It includes a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other through the communication bus; Memory, used to store computer programs; A processor, for implementing the method steps described in any one of claims 1-13 when executing a program stored in a memory.

16. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method steps described in any one of claims 1 to 13 are implemented.

Images