CN116402558A - Bid adjustment method, device, equipment and storage medium for advertisement space - Google Patents
Bid adjustment method, device, equipment and storage medium for advertisement space Download PDFInfo
- Publication number
- CN116402558A CN116402558A CN202310332940.XA CN202310332940A CN116402558A CN 116402558 A CN116402558 A CN 116402558A CN 202310332940 A CN202310332940 A CN 202310332940A CN 116402558 A CN116402558 A CN 116402558A
- Authority
- CN
- China
- Prior art keywords
- user
- quality
- time period
- information
- target media
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 47
- 238000003066 decision tree Methods 0.000 claims description 113
- 230000006870 function Effects 0.000 claims description 41
- 238000013145 classification model Methods 0.000 claims description 36
- 239000004973 liquid crystal related substance Substances 0.000 claims description 24
- 238000004590 computer program Methods 0.000 claims description 15
- 238000012216 screening Methods 0.000 claims description 8
- GOLXNESZZPUPJE-UHFFFAOYSA-N spiromesifen Chemical compound CC1=CC(C)=CC(C)=C1C(C(O1)=O)=C(OC(=O)CC(C)(C)C)C11CCCC1 GOLXNESZZPUPJE-UHFFFAOYSA-N 0.000 claims description 8
- 238000001514 detection method Methods 0.000 claims description 6
- 238000012887 quadratic function Methods 0.000 claims description 6
- 238000010586 diagram Methods 0.000 description 5
- 238000007726 management method Methods 0.000 description 4
- 230000008901 benefit Effects 0.000 description 3
- 230000008569 process Effects 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 230000001360 synchronised effect Effects 0.000 description 2
- 238000013459 approach Methods 0.000 description 1
- 239000000470 constituent Substances 0.000 description 1
- 238000013527 convolutional neural network Methods 0.000 description 1
- 230000010485 coping Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 230000007717 exclusion Effects 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 238000012549 training Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0241—Advertisements
- G06Q30/0273—Determination of fees for advertising
- G06Q30/0275—Auctions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0241—Advertisements
- G06Q30/0242—Determining effectiveness of advertisements
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0241—Advertisements
- G06Q30/0277—Online advertisement
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Landscapes
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Accounting & Taxation (AREA)
- Development Economics (AREA)
- Strategic Management (AREA)
- Finance (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Economics (AREA)
- Marketing (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The application relates to a bid adjustment method, device, equipment and storage medium for advertisement slots. The method comprises the following steps: storing user quality information in a database; acquiring user information of a user in a first time period on a target media advertisement platform, and calculating a high-quality user duty ratio in the first time period; acquiring high-quality user duty cycle data in a second time period; predicting a high quality user duty cycle on the targeted media advertisement platform over a future third time period; and adjusting the bid value of the advertisement space of the target media advertisement platform in real time according to the high-quality user duty ratio prediction result in the future third time period. According to the method and the device for predicting the user quality of the advertisement space, the user quality of the next time period can be accurately predicted, the bid value of the advertisement space is determined according to the user quality, instead of the bid mode of the advertisement space given by adopting a unified strategy, the situation that the bid of the advertisement space is matched with the user quality at all times is achieved, and the matching accuracy of the bid value of the advertisement space and the user quality is improved.
Description
Technical Field
The present disclosure relates to the field of advertisement delivery technologies, and in particular, to a bid adjustment method, apparatus, computer device, and storage medium for an advertisement space.
Background
In the field of advertisement delivery service, RTA refers to real-time screening and bidding intervention of advertisement traffic by advertisers through an API interface. RTA is the abbreviation of real time API, sends the request of the identification of customer identity to the advertiser in the directional link, carries on the customer's screening, is used for meeting the real-time personalized putting demand of the advertiser. RTA service is provided in the system of the advertisement main side to provide an RTA request interface for media, and the media initiates RTA request to the advertisement main side aiming at the advertisement position information to inquire whether a certain user of the advertisement main side participates in the bidding of the advertisement position.
At present, whether a user participates in bidding is judged by an advertisement main side, and whether the user participates in bidding is judged by executing a rule of service configuration by using model scores. Basically, a plurality of fixed policy rules are preset, the value of the model score is output according to advertisement consumption and initial credit amount, and the business personnel can select the executed policy rules according to the value decision of the model score and can only select among the existing fixed policy rules.
However, since the policy rules set in the fixed manner cannot be modified and adjusted in real time, the quality of the current guest group cannot be predicted according to the user quality, and only a unified policy is used for making a decision, so that the bid of the advertisement space given by the RTA policy is not matched with the user quality, the recognition advantage of RTA is weakened, the bid result of the advertisement space provided by the RTA service is finally biased, high-quality user loss is caused, and the occupation ratio of the low-quality user is increased.
Disclosure of Invention
Based on the above, it is necessary to provide a bid adjustment method, device, computer equipment and storage medium for advertisement spots, which can solve the technical problem that the existing RTA service cannot accurately predict the user quality of the current guest group, and the bid of advertisement spots given by the RTA policy is not matched with the user quality.
In one aspect, a method for bid adjustment of an ad spot is provided, the method comprising:
storing user quality information in a database, wherein the user quality information comprises user information and user quality scoring information;
acquiring user information of a user in a first time period on a target media advertisement platform, and calculating a high-quality user duty ratio in the first time period on the target media advertisement platform by combining the user quality information stored in the database;
Acquiring high-quality user duty ratio data in a second time period on a target media advertisement platform, wherein the duration of the second time period is longer than that of the first time period;
comparing and judging the high-quality user duty ratio in the first time period with the high-quality user duty ratio data in the second time period on a target media advertisement platform, and predicting the high-quality user duty ratio in a future third time period on the target media advertisement platform;
and adjusting the bid value of the advertisement position of the target media advertisement platform in real time according to the high-quality user duty ratio prediction result of the target media advertisement platform in a future third time period.
In one embodiment, before the step of storing the user quality information in the database, the method further comprises:
acquiring user information of a user on a target media advertisement platform;
classifying the users according to the user information of the users, and grading each type of users;
summing the user scores corresponding to all the classifications of the user to obtain the quality score of the user;
acquiring quality scores of all users on a target media advertisement platform, arranging all the users according to a descending order of the quality scores, classifying the users according to the rule of 80/20, classifying the first twenty percent of users in the descending order as high-quality users, and classifying the last eighty percent of users in the descending order as low-quality users;
And forming the quality score information of the user by the quality score of the user and the quality classification result of the user.
In one embodiment, the step of obtaining the user information of the user in the first period of time on the target media advertisement platform and calculating the high quality user duty ratio in the first period of time on the target media advertisement platform in combination with the user quality information stored in the database includes:
acquiring user information of a user in a first time period on a target media advertisement platform;
screening out the user information stored in the database from the user information of the user by combining the user quality information stored in the database;
acquiring corresponding quality scoring information according to the screened user information;
and calculating the high-quality user duty ratio of the target media advertisement platform in the first time period according to the acquired user quality scoring information.
In one embodiment, the step of calculating the high quality user duty ratio of the target media advertisement platform in the first time period according to the obtained user quality scoring information includes:
obtaining a classification result of a user in the user quality scoring information;
The statistical classification result is the number A of the users with high quality, and the statistical classification result is the number B of the users with low quality;
and calculating the high-quality user duty ratio of the target media advertisement platform in the first time period according to the formula A/(A+B).
In one embodiment, the step of classifying the users according to the user information of the users and scoring each class of users includes:
acquiring user information of a user, wherein the user information comprises at least one of age, gender, occupation, preference, academic history, income, region, property, family condition, consumption level price, user click data and user watching duration;
constructing a classification model by using an XGBoost algorithm, wherein the classification model comprises a plurality of decision trees;
setting a predicted value of a leaf node at the tail end of each decision tree according to the classification parameters in each decision tree;
classifying the user information of the user into the terminal leaf nodes of the decision tree according to the user information of the user, and taking the predicted value of the terminal leaf nodes of the decision tree as the user score of the user corresponding to the decision tree.
In one embodiment, the xGBoost algorithm is used for constructing a classification model, and the classification model comprises a plurality of decision tree steps, including:
Setting a plurality of classification parameters in each decision tree, wherein the decision tree can classify a user to the end leaf node of the decision tree through the classification parameters;
is provided with
A value representing a total of t decision trees predicted together; />
Representing the predicted value of the 'previous t-1 decision tree' to the sample i; y is i Representing the actual value of sample i; f (f) t (x i ) Representing the predicted value of the t-th decision tree on the sample i; />
Representation and y i ，/>
A related loss function; omega (f) j ) Representing the model complexity of the j-th tree;
the predicted value of the classification model is set as follows:
the loss function of the classification model is set as follows:
wherein (1)>
Representation and y i ，/>
The loss function concerned, i.e. the deviation; />
A complex term representing the j-th tree;
simplifying the loss function:
wherein, the liquid crystal display device comprises a liquid crystal display device,
representing the first t-1 models;
taylor expansion of the loss function:
wherein g i f t (x i ) Is the first derivative of the residual error and,
is the second derivative of the residual;
the parameters of the tree are taken into the loss function:
wherein, the liquid crystal display device comprises a liquid crystal display device,
table omega j Showing the value of the jth leaf node in the tree; t represents the number of leaf nodes; i j Representing samples that are located at the j-th leaf node; />
G representing all samples belonging to the jth leaf node i Sum up; />
Represents h representing all samples belonging to the jth leaf node i Sum up; γt represents a quadratic function.
In one embodiment, the step of adjusting the bid amount of the advertisement space of the target media advertisement platform in real time according to the predicted result of the high-quality user duty ratio in the future third time period on the target media advertisement platform includes:
establishing a bid number association relation between the high-quality user duty ratio and the advertisement position in a database;
acquiring bidding values of corresponding advertisement positions in a database according to a high-quality user duty ratio prediction result in a future third time period on the target media advertisement platform;
and feeding back the bid number of the corresponding advertisement position in the database to the target media advertisement platform as the bid of the advertisement position in a third time period in the future.
In another aspect, an apparatus for bid adjustment of an ad spot is provided, the apparatus comprising:
the database management module is used for storing user quality information in the database, wherein the user quality information comprises user information and user quality scoring information;
the user quality detection module is used for acquiring user information of a user in a first time period on the target media advertisement platform and calculating the high-quality user duty ratio in the first time period on the target media advertisement platform by combining the user quality information stored in the database;
The historical data acquisition module is used for acquiring high-quality user duty ratio data in a second time period on the target media advertisement platform, wherein the duration of the second time period is longer than that of the first time period;
the user quality prediction module is used for comparing and judging the high-quality user duty ratio in the first time period with the high-quality user duty ratio data in the second time period on the target media advertisement platform and predicting the high-quality user duty ratio in a future third time period on the target media advertisement platform;
and the RTA strategy control bidding module is used for adjusting the bidding value of the advertising position of the target media advertising platform in real time according to the high-quality user duty ratio prediction result in the future third time period on the target media advertising platform.
In yet another aspect, a computer device is provided comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, the processor implementing the steps of:
storing user quality information in a database, wherein the user quality information comprises user information and user quality scoring information;
acquiring user information of a user in a first time period on a target media advertisement platform, and calculating a high-quality user duty ratio in the first time period on the target media advertisement platform by combining the user quality information stored in the database;
Acquiring high-quality user duty ratio data in a second time period on a target media advertisement platform, wherein the duration of the second time period is longer than that of the first time period;
comparing and judging the high-quality user duty ratio in the first time period with the high-quality user duty ratio data in the second time period on a target media advertisement platform, and predicting the high-quality user duty ratio in a future third time period on the target media advertisement platform;
and adjusting the bid value of the advertisement position of the target media advertisement platform in real time according to the high-quality user duty ratio prediction result of the target media advertisement platform in a future third time period.
In yet another aspect, a computer readable storage medium is provided, having stored thereon a computer program which when executed by a processor performs the steps of:
storing user quality information in a database, wherein the user quality information comprises user information and user quality scoring information;
acquiring user information of a user in a first time period on a target media advertisement platform, and calculating a high-quality user duty ratio in the first time period on the target media advertisement platform by combining the user quality information stored in the database;
Acquiring high-quality user duty ratio data in a second time period on a target media advertisement platform, wherein the duration of the second time period is longer than that of the first time period;
comparing and judging the high-quality user duty ratio in the first time period with the high-quality user duty ratio data in the second time period on a target media advertisement platform, and predicting the high-quality user duty ratio in a future third time period on the target media advertisement platform;
and adjusting the bid value of the advertisement position of the target media advertisement platform in real time according to the high-quality user duty ratio prediction result of the target media advertisement platform in a future third time period.
According to the bid adjustment method, the bid adjustment device, the computer equipment and the storage medium of the advertisement position, the high-quality user duty ratio of the advertisement position is calculated according to the user information of the user in the current first time period, the real-time high-quality user duty ratio is compared with the historical data, the user quality of the next time period can be accurately predicted, the bid value of the advertisement position is determined according to the predicted and acquired user quality of the next time period instead of the bid mode of the advertisement position given by adopting a unified strategy, the matching of the bid of the advertisement position and the user quality moment is realized, and the matching accuracy of the bid value of the advertisement position and the user quality is improved.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly described below, and it is apparent that the drawings in the following description are only some embodiments of the present invention, and other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is an application environment diagram of a bid adjustment method for ad slots in one embodiment;
FIG. 2 is a flow chart of a bid adjustment method for ad slots in one embodiment;
FIG. 3 is a flow chart illustrating steps prior to the step of storing user quality information in a database in one embodiment;
FIG. 4 is a flowchart illustrating steps for obtaining user information of a user on a target media advertisement platform during a first time period, and calculating a high quality user duty ratio on the target media advertisement platform during the first time period in combination with user quality information stored in the database according to one embodiment;
FIG. 5 is a flowchart illustrating steps for calculating a high quality user duty cycle for a first time period on the targeted media advertisement platform based on the obtained user quality score information, according to one embodiment;
FIG. 6 is a flow chart of the steps of classifying users according to user information of the users and scoring each type of users in one embodiment;
FIG. 7 is a flow chart of a user scoring assignment step for each class of users of a decision tree using decision tree approach to categorize users based on user information in one embodiment;
FIG. 8 is a flowchart illustrating a step of adjusting the bid amount for an ad spot on the target media advertisement platform in real time based on the high quality user duty prediction result in a third time period in the future on the target media advertisement platform in one embodiment;
FIG. 9 is a schematic diagram of an application of a bid adjustment method for ad slots in one embodiment;
FIG. 10 is a block diagram of an apparatus for bid adjustment of ad slots in one embodiment;
FIG. 11 is an internal block diagram of a computer device in one embodiment.
Detailed Description
In order to make the objects, technical solutions and advantages of the present application more apparent, the present application will be further described in detail with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the present application.
As described in the background art, a general advertiser uses RTA service, only uses a unified strategy to make decisions, and the bid of an advertisement space given by the RTA strategy is not matched with the user quality according to the quality mode of the user quality prediction current guest group, so that the bid result of the advertisement space provided by the RTA service is deviated, high-quality user loss is caused, and the occupation ratio of a low-quality user is increased. Moreover, the bidding mode of the advertisement space given by the RTA service using the unified strategy for making decisions is difficult to realize flexible and changeable coping strategies, for example, high-quality users on weekends or holidays are increased, so that the ratio of the high-quality users on weekdays is obviously different from that of the high-quality users on weekdays, and the bidding mode of the advertisement space given by the RTA service using the unified strategy for making decisions is difficult to meet the demands of the users.
Therefore, the embodiment of the invention provides a bid adjustment method for advertisement space, in actual advertisement delivery, the bid adjustment method for advertisement space is found to be capable of predicting the quality level of the current guest group according to the duty ratio of high-quality users in advertisement time slots (for example, 05:00-09:00) in certain time slots of the current day, and then different RTA exclusion strategies are executed according to the guest group quality, so that a proper bid mode of the advertisement space is selected.
Specifically, the quality score is calculated by a model based on the user representation (i.e., user information) such as age, academic, income, region, etc. And then comparing and judging the user duty ratio of the real-time high quality score with the historical data through daily exposure, clicking and registering the quality score of the user checked by the user, and predicting the user quality of the next period. Thereby adjusting the executed strategy and realizing the matching of the bidding of the advertisement space and the quality of the user.
The bid adjustment method for the advertisement space can be applied to an application environment shown in fig. 1. Wherein the terminal 102 communicates with the media advertisement platform 103 via a network, and the media advertisement platform 103 communicates with the advertiser server 104 via a network. The terminal 102 may be, but not limited to, various personal computers, notebook computers, smartphones, tablet computers, and portable wearable devices, and the client may browse advertisement information such as headlines, vacations, etc. through the terminal 102, the media advertisement platform 103 may include servers such as headlines, vacations, etc., and the media advertisement platform 103 and the advertiser server 104 may be implemented by separate servers or a server cluster formed by a plurality of servers.
It is appreciated that the user population characteristics of any one of the media advertisement platforms 103 remain substantially unchanged, and any one of the media advertisement platforms 103 may be considered a targeted media advertisement platform.
In one embodiment, as shown in fig. 2, a bid adjustment method for an advertisement slot is provided, and the method is applied to the advertiser server 104 in fig. 1 for illustration, and includes the following steps:
step S1, storing user quality information in a database, wherein the user quality information comprises user information and user quality scoring information;
step S2, obtaining user information of a user in a first time period on a target media advertisement platform, and calculating a high-quality user duty ratio in the first time period on the target media advertisement platform by combining the user quality information stored in the database;
step S3, obtaining high-quality user duty ratio data in a second time period on the target media advertisement platform, wherein the duration of the second time period is longer than that of the first time period;
s4, comparing and judging the high-quality user duty ratio in the first time period with the high-quality user duty ratio data in the second time period on the target media advertisement platform, and predicting the high-quality user duty ratio in a future third time period on the target media advertisement platform;
And S5, adjusting the bid value of the advertisement position of the target media advertisement platform in real time according to the high-quality user duty ratio prediction result in the future third time period on the target media advertisement platform.
It can be appreciated that the present application utilizes the high quality user duty cycle data on the target media advertisement platform during the second time period to evaluate the high quality user duty cycle of the user during the first time period on the target media advertisement platform, and predicts the high quality user duty cycle during the third time period in the future based on the high quality user duty cycle evaluation result, thereby giving the estimated bid value of the advertisement space.
The duration of the second time period is longer than that of the first time period, so that the high-quality user duty ratio prediction result in a future third time period after the first time period can be obtained based on the high-quality user duty ratio statistical curve in the second time period by finding out the high-quality user duty ratio matched or similar to the first time period in the second time period.
In this embodiment, an RTA service is provided to the media advertisement platform by means of an API interface; and the media advertisement platform generates an idle advertisement position according to the process that a client browses a webpage, sends an advertisement resource request corresponding to the idle advertisement position, and initiates RTA consultation through the API interface when processing the advertisement resource request. The bid adjustment method of the advertisement space is used for replying the RTA consultation to the bid value of the advertisement space.
In order to obtain the bidding range of the corresponding advertisement space of the target media advertisement platform, data statistics is required to be performed on the user of the target media advertisement platform before the user quality information is stored in the database, so as to obtain the quality level or quality distribution of the user of the target media advertisement platform.
Thus, as shown in fig. 3, before the step of storing the user quality information in the database, the method further includes:
step S11, obtaining user information of a user on a target media advertisement platform;
step S12, classifying the using users according to the user information of the using users, and grading each type of users;
step S13, summing the user scores corresponding to all the classifications of the user to obtain the quality score of the user;
step S14, obtaining quality scores of all users on a target media advertisement platform, arranging all the users according to a descending order of the quality scores, classifying the users according to the rule of 80/20, classifying the first twenty percent of users in the descending order as high-quality users, and classifying the last eighty percent of users in the descending order as low-quality users;
and S15, forming user quality scoring information by the quality scoring of the user and the quality classification result of the user.
As shown in fig. 4, the step of obtaining the user information of the user in the first period of time on the target media advertisement platform and calculating the high quality user duty ratio in the first period of time on the target media advertisement platform by combining the user quality information stored in the database includes:
step S21, obtaining user information of a user in a first time period on a target media advertisement platform;
step S22, screening out the user information stored in the database from the user information of the user by combining the user quality information stored in the database;
step S23, corresponding quality grading information is obtained according to the screened user information;
and step S24, calculating the high-quality user duty ratio of the target media advertisement platform in the first time period according to the acquired user quality scoring information.
In this embodiment, the user in the first period is used to find out whether the user is stored in the user quality information in the database, if yes, the user is taken as a judgment factor, and if not, the user is directly rejected. And finding out the user quality information stored in the database from the using users in the first time period, acquiring corresponding quality scoring information according to the user quality information, and then calculating the corresponding high-quality user duty ratio.
The rejected users in the first time period can be used as data sources for subsequent updating of the database information, the user information of the users is collected and subjected to quality scoring and quality classification to obtain corresponding quality scoring information, the database can be updated to perfect the quantity of the user quality information stored in the database, and the comprehensive coverage of the user quality information of the database is improved. The method specifically comprises the following steps: and obtaining user information which is not stored in the database in the user information of the user, carrying out quality grading and quality classification on the user information which is not stored in the database to obtain corresponding quality grading information, and storing the user information which is not stored in the database and the corresponding quality grading information into the database.
As shown in fig. 5, the step of calculating the high quality user duty ratio in the first time period on the target media advertisement platform according to the obtained user quality score information includes:
step S31, obtaining a classification result of a user in the user quality scoring information;
step S32, counting the number A of the users with high quality as the classification result, and counting the number B of the users with low quality as the classification result;
And step S33, calculating the high-quality user duty ratio of the target media advertisement platform in the first time period according to the formula A/(A+B).
As shown in fig. 6, the step of classifying the users according to the user information of the users and scoring each type of users includes:
step S41, obtaining user information of a user, wherein the user information comprises at least one of age, gender, occupation, preference, academic, income, region, property, family condition, consumption level price, user click data and user watching duration;
and S42, classifying the using users according to the user information by adopting a decision tree mode, and carrying out user scoring assignment on each type of using users of the decision tree.
Referring to fig. 9, the user information includes user click data, user exposure data, and user registration data, wherein the user exposure data is a user viewing time period, and the user registration data includes age, gender, occupation, preference, academic, income, region, property, family condition, and consumption level price. The consumption level price can be obtained by obtaining the corresponding mobile phone price through the mobile phone model.
As shown in fig. 7, the step of classifying the users according to the user information by using a decision tree mode and performing user scoring assignment on each type of users of the decision tree includes:
Step S51, constructing a classification model by using an XGBoost algorithm, wherein the classification model comprises a plurality of decision trees; each decision tree comprises a plurality of classification parameters through which users can be classified to the end leaf nodes of the decision tree;
step S52, according to the classification parameters in each decision tree, setting the predicted value of the leaf node at the tail end of the decision tree;
step S53, classifying the user information of the user into the terminal leaf nodes of the decision tree, and taking the predicted value of the terminal leaf nodes of the decision tree as the user score of the user corresponding to the decision tree.
That is, the step of classifying the using users according to the user information of the using users and scoring each class of users includes step S41, step S51, step S52, step S53.
XGBoost is an integrated learning method, is totally called eXtreme Gradient Boosting, is an implementation of a gradient lifting decision tree algorithm, and is the fastest and best open source boosting tree tool kit at present, and the speed is more than 10 times faster than that of a common tool kit.
The ensemble learning (Ensemble Learning) is a model formed by fusing a plurality of weak models (base models), and the potential idea is that even if one weak model is mispredicted, the other weak models can correct the errors back. There are generally two kinds of fusion strategies, bagging and Boosting, XGBoost is an excellent representation of Boosting ensemble learning.
XGBoost has the advantages of very strong fitting ability, almost the strongest in the traditional machine learning algorithm, more super-parameters and greater difficulty in understanding and proficiency in parameter adjustment.
In practice, a large number of models are XGBoost algorithms, so that the most commonly used model algorithms are not known.
The XGBoost algorithm is a scalable Tree boosting algorithm. The basic constituent elements of XGboost are: a decision tree; we turn these decision trees into "weak learners" that together make up XGboost. The data set used to generate each decision tree is the entire data set. The generation of each decision tree can be considered a complete decision tree generation process.
The "weak learner" of XGboost is a "decision tree", each of which is a model when the objective function value is minimum. Only if the objective function value of this "decision tree" is minimal will it be selected as a "weak learner".
Total t tree predictions for sample i = previous t-1 predictions + t-th predictions.
The XGBoost algorithm used is used for constructing a classification model, the classification model comprises a plurality of decision tree steps, and the XGBoost algorithm comprises the following steps:
Setting a plurality of classification parameters in each decision tree, wherein the decision tree can classify a user to the end leaf node of the decision tree through the classification parameters;
is provided with
A value representing a total of t decision trees predicted together; />
Representing the predicted value of the 'previous t-1 decision tree' to the sample i; y is i Representing the actual value of sample i; f (f) t (x i ) Representing the predicted value of the t-th decision tree on the sample i; />
Representation and y i ，/>
A related loss function; omega (f) j ) Representing the model complexity of the j-th tree;
the predicted value of the classification model is set as follows:
the loss function of the classification model is set as follows:
wherein (1)>
Representation and y i ，/>
The loss function concerned, i.e. the deviation; />
A complex term representing the j-th tree;
simplifying the loss function:
wherein, the liquid crystal display device comprises a liquid crystal display device,
representing the first t-1 models;
taylor expansion of the loss function:
wherein g i f t (x i ) Is the first derivative of the residual error and,
is the second derivative of the residual;
the parameters of the tree are taken into the loss function:
wherein, the liquid crystal display device comprises a liquid crystal display device,
table omega j Showing the value of the jth leaf node in the tree; t represents the number of leaf nodes; i j Representing samples that are located at the j-th leaf node; />
G representing all samples belonging to the jth leaf node i Sum up; />
Represents h representing all samples belonging to the jth leaf node i Sum up; γt represents a quadratic function.
Is provided with
Then the value ω of the leaf node is set at the time of setting the value of the leaf node j Is set as
The objective function of the tree is at this point minimal.
The values of the leaf nodes and the size of the objective function are related to the bias of the "top k-1 decision trees". And each decision tree has a minimum objective function on the premise of structure determination
That is: the first and second derivatives of the deviation of the samples in each leaf node are divided and then summed for all leaf nodes.
As shown in fig. 8, the step of adjusting the bid amount of the advertisement space of the target media advertisement platform in real time according to the predicted result of the high-quality user duty ratio in the future third time period on the target media advertisement platform includes:
step S61, establishing a bid number association relation between the high-quality user duty ratio and the advertisement position in a database;
step S62, obtaining bidding values of corresponding advertisement positions in a database according to a high-quality user duty ratio prediction result in a future third time period on the target media advertisement platform;
and step S63, feeding back the bid value of the corresponding advertisement position in the database to the target media advertisement platform as the bid of the advertisement position in the future third time period.
That is, the bid amount of the ad spot is determined based on a predetermined amount corresponding to the high quality user's duty cycle. And comparing and judging the real-time high-quality user duty ratio with the historical data, and predicting the user quality of the next time period, so as to adjust the bid value of the advertisement space. The price adjustment curve can be formed by continuously adjusting the bid values of the advertisement slots in a plurality of time periods, so that the price adjustment curve is matched with the user quality of each time period, and when the high-quality user in a certain time period has low duty ratio, the price cannot be given out, and the low-quality user is prevented from being selected. When the high quality user ratio is high for a certain period of time, a high price is given to rob the high quality user.
As shown in fig. 9, the second time period is preferably approximately 7 days, and the cycle law with the duration of each week is known according to the high quality user duty ratio of approximately 7 days. Of course the second period of time may also be one month. In this embodiment, the first time period and the third time period are preferably half an hour, so that the bid number of the advertisement space can be timely adjusted, and the situation of processing data increase caused by frequent adjustment of the bid of the advertisement space can be avoided. It is understood that the first time period and the third time period are not necessarily the same, and the first time period and the third time period may be other durations such as one hour and two hours.
According to the bid adjustment method of the advertisement space, the high-quality user duty ratio is calculated according to the user information of the user in the current first time period, the real-time high-quality user duty ratio is compared with the historical data, the user quality of the next time period can be accurately predicted, the bid value of the advertisement space is determined according to the predicted user quality of the next time period instead of the bid mode of the advertisement space given by adopting a unified strategy, the situation that the bid of the advertisement space is matched with the user quality moment is achieved, and the matching accuracy of the bid value of the advertisement space and the user quality is improved.
It should be understood that, although the steps in the flowcharts of fig. 2-9 are shown in order as indicated by the arrows, these steps are not necessarily performed in order as indicated by the arrows. The steps are not strictly limited to the order of execution unless explicitly recited herein, and the steps may be executed in other orders. Moreover, at least some of the steps of fig. 2-9 may include multiple sub-steps or stages that are not necessarily performed at the same time, but may be performed at different times, nor does the order in which the sub-steps or stages are performed necessarily occur in sequence, but may be performed alternately or alternately with at least a portion of other steps or sub-steps or stages of other steps.
In one embodiment, as shown in FIG. 10, an apparatus 10 for bid adjustment of an ad spot is provided, comprising: a database management module 1, a user quality detection module 2, a historical data acquisition module 3, a user quality prediction module 4 and an RTA strategy control bidding module 5.
The database management module 1 is configured to store user quality information in a database, where the user quality information includes user information and user quality score information.
The user quality detection module 2 (corresponding to the AI model in fig. 9) is configured to obtain user information of a user on the target media advertisement platform during a first period, and calculate a high quality user duty cycle on the target media advertisement platform during the first period in combination with the user quality information stored in the database. The user quality detection module 2 can obtain a converged convolutional neural network (AI) model after training, so as to automatically obtain user information of a user in a first time period on the target media advertisement platform, and calculate a corresponding high-quality user duty ratio.
The historical data acquisition module 3 is configured to acquire high-quality user duty data on the target media advertisement platform in a second period of time, where the duration of the second period of time is greater than the duration of the first period of time.
The user quality prediction module 4 (corresponding to the prediction model in fig. 9) is configured to compare the high quality user duty ratio in the first time period with the high quality user duty ratio data in the second time period on the target media advertisement platform, and predict the high quality user duty ratio in the third time period in the future on the target media advertisement platform. The user quality prediction module 4, i.e. the prediction model, is configured to predict the high quality user duty cycle in a third time period in the future based on the high quality user duty cycle data in the second time period according to the high quality user duty cycle in the first time period on the targeted media advertisement platform.
The RTA policy control bidding module 5 is configured to adjust a bid value of an advertisement slot of the target media advertisement platform in real time according to a high quality user duty prediction result of the target media advertisement platform in a third time period.
Wherein, before the step of storing the user quality information in the database, the database management module 1 further includes:
acquiring user information of a user on a target media advertisement platform;
classifying the users according to the user information of the users, and grading each type of users;
Summing the user scores corresponding to all the classifications of the user to obtain the quality score of the user;
acquiring quality scores of all users on a target media advertisement platform, arranging all the users according to a descending order of the quality scores, classifying the users according to the rule of 80/20, classifying the first twenty percent of users in the descending order as high-quality users, and classifying the last eighty percent of users in the descending order as low-quality users;
and forming the quality score information of the user by the quality score of the user and the quality classification result of the user.
The user quality detection module 2 (corresponding to the AI model in fig. 9) is configured to obtain user information of a user on a target media advertisement platform in a first period, and calculate a high quality user duty ratio on the target media advertisement platform in the first period by combining the user quality information stored in the database, where the method includes:
acquiring user information of a user in a first time period on a target media advertisement platform;
screening out the user information stored in the database from the user information of the user by combining the user quality information stored in the database;
Acquiring corresponding quality scoring information according to the screened user information;
and calculating the high-quality user duty ratio of the target media advertisement platform in the first time period according to the acquired user quality scoring information.
In this embodiment, the user in the first period is used to find out whether the user is stored in the user quality information in the database, if yes, the user is taken as a judgment factor, and if not, the user is directly rejected. And finding out the user quality information stored in the database from the using users in the first time period, acquiring corresponding quality scoring information according to the user quality information, and then calculating the corresponding high-quality user duty ratio.
The rejected users in the first time period can be used as data sources for subsequent updating of the database information, the user information of the users is collected and subjected to quality scoring and quality classification to obtain corresponding quality scoring information, the database can be updated to perfect the quantity of the user quality information stored in the database, and the comprehensive coverage of the user quality information of the database is improved. The method specifically comprises the following steps: and obtaining user information which is not stored in the database in the user information of the user, carrying out quality grading and quality classification on the user information which is not stored in the database to obtain corresponding quality grading information, and storing the user information which is not stored in the database and the corresponding quality grading information into the database.
The step of calculating the high-quality user duty ratio of the target media advertisement platform in the first time period according to the obtained user quality scoring information comprises the following steps:
obtaining a classification result of a user in the user quality scoring information;
the statistical classification result is the number A of the users with high quality, and the statistical classification result is the number B of the users with low quality;
and calculating the high-quality user duty ratio of the target media advertisement platform in the first time period according to the formula A/(A+B).
The step of classifying the users according to the user information of the users and grading each type of users comprises the following steps:
acquiring user information of a user, wherein the user information comprises at least one of age, gender, occupation, preference, academic history, income, region, property, family condition, consumption level price, user click data and user watching duration;
classifying the users according to the user information by adopting a decision tree mode, and carrying out user scoring assignment on each type of users of the decision tree.
The user information comprises user click data, user exposure data and user registration data, wherein the user exposure data is user watching time, and the user registration data comprises age, gender, occupation, preference, academic, income, territory, property, family condition and consumption level price. The consumption level price can be obtained by obtaining the corresponding mobile phone price through the mobile phone model.
The step of classifying the using users according to the user information by adopting a decision tree mode and carrying out user scoring and assignment on each type of using users of the decision tree comprises the following steps:
constructing a classification model by using an XGBoost algorithm, wherein the classification model comprises a plurality of decision trees; each decision tree comprises a plurality of classification parameters through which users can be classified to the end leaf nodes of the decision tree;
setting a predicted value of a leaf node at the tail end of each decision tree according to the classification parameters in each decision tree;
classifying the user information of the user into the terminal leaf nodes of the decision tree according to the user information of the user, and taking the predicted value of the terminal leaf nodes of the decision tree as the user score of the user corresponding to the decision tree.
The XGBoost algorithm used is used for constructing a classification model, the classification model comprises a plurality of decision tree steps, and the XGBoost algorithm comprises the following steps:
setting a plurality of classification parameters in each decision tree, wherein the decision tree can classify a user to the end leaf node of the decision tree through the classification parameters;
is provided with
A value representing a total of t decision trees predicted together; />
Representing the predicted value of the 'previous t-1 decision tree' to the sample i; y is i Representing the actual value of sample i; f (f) t (x i ) Representing the predicted value of the t-th decision tree on the sample i; />
Representation and y i ，/>
A related loss function; omega (f) j ) Representing the model complexity of the j-th tree;
the predicted value of the classification model is set as follows:
the loss function of the classification model is set as follows:
wherein (1)>
Representation and y i ，/>
The loss function concerned, i.e. the deviation; />
A complex term representing the j-th tree;
simplifying the loss function:
wherein, the liquid crystal display device comprises a liquid crystal display device,
representing the first t-1 models;
taylor expansion of the loss function:
wherein g i f t (x i ) Is the first derivative of the residual error and,
is the second derivative of the residual;
the parameters of the tree are taken into the loss function:
wherein, the liquid crystal display device comprises a liquid crystal display device,
table omega j Showing the value of the jth leaf node in the tree; t represents the number of leaf nodes; i j Representing samples that are located at the j-th leaf node; />
G representing all samples belonging to the jth leaf node i Sum up; />
Represents h representing all samples belonging to the jth leaf node i Sum up; γt represents a quadratic function.
Wherein, the RTA policy control bidding module 5 is configured to adjust, in real time, a bid amount of an advertisement space of the target media advertisement platform according to a high-quality user duty prediction result in a third time period in the future on the target media advertisement platform, including:
Establishing a bid number association relation between the high-quality user duty ratio and the advertisement position in a database;
acquiring bidding values of corresponding advertisement positions in a database according to a high-quality user duty ratio prediction result in a future third time period on the target media advertisement platform;
and feeding back the bid number of the corresponding advertisement position in the database to the target media advertisement platform as the bid of the advertisement position in a third time period in the future.
For specific limitations on the means of bid adjustment of an ad spot, reference may be made to the limitations of the bid adjustment method for an ad spot hereinabove, and no further description is given here. The various modules in the above-described apparatus for bid adjustment of ad slots may be implemented in whole or in part by software, hardware, and combinations thereof. The above modules may be embedded in hardware or may be independent of a processor in the computer device, or may be stored in software in a memory in the computer device, so that the processor may call and execute operations corresponding to the above modules.
In one embodiment, a computer device is provided, which may be a server, and the internal structure of which may be as shown in fig. 11. The computer device includes a processor, a memory, a network interface, and a database connected by a system bus. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The database of the computer device is for storing data of bid adjustments for ad slots. The network interface of the computer device is used for communicating with an external terminal through a network connection. The computer program when executed by a processor implements a method of bid adjustment for ad slots.
It will be appreciated by those skilled in the art that the structure shown in fig. 11 is merely a block diagram of a portion of the structure associated with the present application and is not limiting of the computer device to which the present application applies, and that a particular computer device may include more or fewer components than shown, or may combine some of the components, or have a different arrangement of components.
In one embodiment, a computer device is provided comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, the processor implementing the steps of when executing the computer program:
storing user quality information in a database, wherein the user quality information comprises user information and user quality scoring information;
acquiring user information of a user in a first time period on a target media advertisement platform, and calculating a high-quality user duty ratio in the first time period on the target media advertisement platform by combining the user quality information stored in the database;
acquiring high-quality user duty ratio data in a second time period on a target media advertisement platform, wherein the duration of the second time period is longer than that of the first time period;
Comparing and judging the high-quality user duty ratio in the first time period with the high-quality user duty ratio data in the second time period on a target media advertisement platform, and predicting the high-quality user duty ratio in a future third time period on the target media advertisement platform;
and adjusting the bid value of the advertisement position of the target media advertisement platform in real time according to the high-quality user duty ratio prediction result of the target media advertisement platform in a future third time period.
Before the step of storing the user quality information in the database, the method further comprises the following steps:
acquiring user information of a user on a target media advertisement platform;
classifying the users according to the user information of the users, and grading each type of users;
summing the user scores corresponding to all the classifications of the user to obtain the quality score of the user;
acquiring quality scores of all users on a target media advertisement platform, arranging all the users according to a descending order of the quality scores, classifying the users according to the rule of 80/20, classifying the first twenty percent of users in the descending order as high-quality users, and classifying the last eighty percent of users in the descending order as low-quality users;
And forming the quality score information of the user by the quality score of the user and the quality classification result of the user.
The step of obtaining the user information of the user in the first time period on the target media advertisement platform and calculating the high-quality user duty ratio in the first time period on the target media advertisement platform by combining the user quality information stored in the database comprises the following steps:
acquiring user information of a user in a first time period on a target media advertisement platform;
screening out the user information stored in the database from the user information of the user by combining the user quality information stored in the database;
acquiring corresponding quality scoring information according to the screened user information;
and calculating the high-quality user duty ratio of the target media advertisement platform in the first time period according to the acquired user quality scoring information.
In this embodiment, the user in the first period is used to find out whether the user is stored in the user quality information in the database, if yes, the user is taken as a judgment factor, and if not, the user is directly rejected. And finding out the user quality information stored in the database from the using users in the first time period, acquiring corresponding quality scoring information according to the user quality information, and then calculating the corresponding high-quality user duty ratio.
The rejected users in the first time period can be used as data sources for subsequent updating of the database information, the user information of the users is collected and subjected to quality scoring and quality classification to obtain corresponding quality scoring information, the database can be updated to perfect the quantity of the user quality information stored in the database, and the comprehensive coverage of the user quality information of the database is improved. The method specifically comprises the following steps: and obtaining user information which is not stored in the database in the user information of the user, carrying out quality grading and quality classification on the user information which is not stored in the database to obtain corresponding quality grading information, and storing the user information which is not stored in the database and the corresponding quality grading information into the database.
The step of calculating the high-quality user duty ratio of the target media advertisement platform in the first time period according to the obtained user quality scoring information comprises the following steps:
obtaining a classification result of a user in the user quality scoring information;
the statistical classification result is the number A of the users with high quality, and the statistical classification result is the number B of the users with low quality;
And calculating the high-quality user duty ratio of the target media advertisement platform in the first time period according to the formula A/(A+B).
The step of classifying the users according to the user information of the users and grading each type of users comprises the following steps:
acquiring user information of a user, wherein the user information comprises at least one of age, gender, occupation, preference, academic history, income, region, property, family condition, consumption level price, user click data and user watching duration;
classifying the users according to the user information by adopting a decision tree mode, and carrying out user scoring assignment on each type of users of the decision tree.
The user information comprises user click data, user exposure data and user registration data, wherein the user exposure data is user watching time, and the user registration data comprises age, gender, occupation, preference, academic, income, territory, property, family condition and consumption level price. The consumption level price can be obtained by obtaining the corresponding mobile phone price through the mobile phone model.
The step of classifying the using users according to the user information by adopting a decision tree mode and carrying out user scoring and assignment on each type of using users of the decision tree comprises the following steps:
Constructing a classification model by using an XGBoost algorithm, wherein the classification model comprises a plurality of decision trees; each decision tree comprises a plurality of classification parameters through which users can be classified to the end leaf nodes of the decision tree;
setting a predicted value of a leaf node at the tail end of each decision tree according to the classification parameters in each decision tree;
classifying the user information of the user into the terminal leaf nodes of the decision tree according to the user information of the user, and taking the predicted value of the terminal leaf nodes of the decision tree as the user score of the user corresponding to the decision tree.
The XGBoost algorithm used is used for constructing a classification model, the classification model comprises a plurality of decision tree steps, and the XGBoost algorithm comprises the following steps:
setting a plurality of classification parameters in each decision tree, wherein the decision tree can classify a user to the end leaf node of the decision tree through the classification parameters;
is provided with
A value representing a total of t decision trees predicted together;/>
representing the predicted value of the 'previous t-1 decision tree' to the sample i; y is i Representing the actual value of sample i; f (f) t (x i ) Representing the predicted value of the t-th decision tree on the sample i; />
Representation and y i ，/>
A related loss function; omega (f) j ) Representing the model complexity of the j-th tree;
The predicted value of the classification model is set as follows:
the loss function of the classification model is set as follows:
wherein (1)>
Representation and y i ，/>
The loss function concerned, i.e. the deviation; />
A complex term representing the j-th tree;
simplifying the loss function:
wherein, the liquid crystal display device comprises a liquid crystal display device,
representing the first t-1 models;
taylor expansion of the loss function:
wherein g i f t (x i ) Is the first derivative of the residual error and,
is the second derivative of the residual;
the parameters of the tree are taken into the loss function:
wherein, the liquid crystal display device comprises a liquid crystal display device,
table omega j Showing the value of the jth leaf node in the tree; t represents the number of leaf nodes; i j Representing samples that are located at the j-th leaf node; />
G representing all samples belonging to the jth leaf node i Sum up; />
Represents h representing all samples belonging to the jth leaf node i Sum up; γt represents a quadratic function.
Specific limitations regarding implementation steps of the processor when executing the computer program may be found in the above limitations of the bid adjustment method for ad slots, and will not be described in detail herein.
In one embodiment, a computer readable storage medium is provided having a computer program stored thereon, which when executed by a processor, performs the steps of:
storing user quality information in a database, wherein the user quality information comprises user information and user quality scoring information;
Acquiring user information of a user in a first time period on a target media advertisement platform, and calculating a high-quality user duty ratio in the first time period on the target media advertisement platform by combining the user quality information stored in the database;
acquiring high-quality user duty ratio data in a second time period on a target media advertisement platform, wherein the duration of the second time period is longer than that of the first time period;
comparing and judging the high-quality user duty ratio in the first time period with the high-quality user duty ratio data in the second time period on a target media advertisement platform, and predicting the high-quality user duty ratio in a future third time period on the target media advertisement platform;
and adjusting the bid value of the advertisement position of the target media advertisement platform in real time according to the high-quality user duty ratio prediction result of the target media advertisement platform in a future third time period.
Before the step of storing the user quality information in the database, the method further comprises the following steps:
acquiring user information of a user on a target media advertisement platform;
classifying the users according to the user information of the users, and grading each type of users;
Summing the user scores corresponding to all the classifications of the user to obtain the quality score of the user;
acquiring quality scores of all users on a target media advertisement platform, arranging all the users according to a descending order of the quality scores, classifying the users according to the rule of 80/20, classifying the first twenty percent of users in the descending order as high-quality users, and classifying the last eighty percent of users in the descending order as low-quality users;
and forming the quality score information of the user by the quality score of the user and the quality classification result of the user.
The step of obtaining the user information of the user in the first time period on the target media advertisement platform and calculating the high-quality user duty ratio in the first time period on the target media advertisement platform by combining the user quality information stored in the database comprises the following steps:
acquiring user information of a user in a first time period on a target media advertisement platform;
screening out the user information stored in the database from the user information of the user by combining the user quality information stored in the database;
Acquiring corresponding quality scoring information according to the screened user information;
and calculating the high-quality user duty ratio of the target media advertisement platform in the first time period according to the acquired user quality scoring information.
In this embodiment, the user in the first period is used to find out whether the user is stored in the user quality information in the database, if yes, the user is taken as a judgment factor, and if not, the user is directly rejected. And finding out the user quality information stored in the database from the using users in the first time period, acquiring corresponding quality scoring information according to the user quality information, and then calculating the corresponding high-quality user duty ratio.
The rejected users in the first time period can be used as data sources for subsequent updating of the database information, the user information of the users is collected and subjected to quality scoring and quality classification to obtain corresponding quality scoring information, the database can be updated to perfect the quantity of the user quality information stored in the database, and the comprehensive coverage of the user quality information of the database is improved. The method specifically comprises the following steps: and obtaining user information which is not stored in the database in the user information of the user, carrying out quality grading and quality classification on the user information which is not stored in the database to obtain corresponding quality grading information, and storing the user information which is not stored in the database and the corresponding quality grading information into the database.
The step of calculating the high-quality user duty ratio of the target media advertisement platform in the first time period according to the obtained user quality scoring information comprises the following steps:
obtaining a classification result of a user in the user quality scoring information;
the statistical classification result is the number A of the users with high quality, and the statistical classification result is the number B of the users with low quality;
and calculating the high-quality user duty ratio of the target media advertisement platform in the first time period according to the formula A/(A+B).
The step of classifying the users according to the user information of the users and grading each type of users comprises the following steps:
acquiring user information of a user, wherein the user information comprises at least one of age, gender, occupation, preference, academic history, income, region, property, family condition, consumption level price, user click data and user watching duration;
classifying the users according to the user information by adopting a decision tree mode, and carrying out user scoring assignment on each type of users of the decision tree.
The user information comprises user click data, user exposure data and user registration data, wherein the user exposure data is user watching time, and the user registration data comprises age, gender, occupation, preference, academic, income, territory, property, family condition and consumption level price. The consumption level price can be obtained by obtaining the corresponding mobile phone price through the mobile phone model.
The step of classifying the using users according to the user information by adopting a decision tree mode and carrying out user scoring and assignment on each type of using users of the decision tree comprises the following steps:
constructing a classification model by using an XGBoost algorithm, wherein the classification model comprises a plurality of decision trees; each decision tree comprises a plurality of classification parameters through which users can be classified to the end leaf nodes of the decision tree;
setting a predicted value of a leaf node at the tail end of each decision tree according to the classification parameters in each decision tree;
classifying the user information of the user into the terminal leaf nodes of the decision tree according to the user information of the user, and taking the predicted value of the terminal leaf nodes of the decision tree as the user score of the user corresponding to the decision tree.
The XGBoost algorithm used is used for constructing a classification model, the classification model comprises a plurality of decision tree steps, and the XGBoost algorithm comprises the following steps:
setting a plurality of classification parameters in each decision tree, wherein the decision tree can classify a user to the end leaf node of the decision tree through the classification parameters;
is provided with
A value representing a total of t decision trees predicted together; />
Representing the predicted value of the 'previous t-1 decision tree' to the sample i; y is i Representing the actual value of sample i; f (f) t (x i ) Representing the predicted value of the t-th decision tree on the sample i; />
Representation and y i ，/>
A related loss function; omega (f) j ) Representing the model complexity of the j-th tree;
the predicted value of the classification model is set as follows:
the loss function of the classification model is set as follows:
wherein (1)>
Representation and y i ，/>
The loss function concerned, i.e. the deviation; />
A complex term representing the j-th tree;
simplifying the loss function:
wherein, the liquid crystal display device comprises a liquid crystal display device,
representing the first t-1 models; />
Taylor expansion of the loss function:
wherein g i f t (x i ) Is the first derivative of the residual error and,
is the second derivative of the residual;
the parameters of the tree are taken into the loss function:
wherein, the liquid crystal display device comprises a liquid crystal display device,
table omega j Showing the value of the jth leaf node in the tree; t represents the number of leaf nodes; i j Representing samples that are located at the j-th leaf node; />
G representing all samples belonging to the jth leaf node i Sum up; />
Represents h representing all samples belonging to the jth leaf node i Sum up; γt represents a quadratic function.
For specific limitations regarding implementation steps of the computer program when executed by the processor, reference may be made to the limitations of the bid adjustment method for ad slots hereinabove, and no further description is given here.
Those skilled in the art will appreciate that implementing all or part of the above described methods may be accomplished by way of a computer program stored on a non-transitory computer readable storage medium, which when executed, may comprise the steps of the embodiments of the methods described above. Any reference to memory, storage, database, or other medium used in the various embodiments provided herein may include non-volatile and/or volatile memory. The nonvolatile memory can include Read Only Memory (ROM), programmable ROM (PROM), electrically Programmable ROM (EPROM), electrically Erasable Programmable ROM (EEPROM), or flash memory. Volatile memory can include Random Access Memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms such as Static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double Data Rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous Link DRAM (SLDRAM), memory bus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), among others.
The technical features of the above embodiments may be arbitrarily combined, and all possible combinations of the technical features in the above embodiments are not described for brevity of description, however, as long as there is no contradiction between the combinations of the technical features, they should be considered as the scope of the description.
The above examples merely represent a few embodiments of the present application, which are described in more detail and are not to be construed as limiting the scope of the invention. It should be noted that it would be apparent to those skilled in the art that various modifications and improvements could be made without departing from the spirit of the present application, which would be within the scope of the present application. Accordingly, the scope of protection of the present application is to be determined by the claims appended hereto.
Claims (10)
1. A bid adjustment method for an advertisement space, comprising:
storing user quality information in a database, wherein the user quality information comprises user information and user quality scoring information;
acquiring user information of a user in a first time period on a target media advertisement platform, and calculating a high-quality user duty ratio in the first time period on the target media advertisement platform by combining the user quality information stored in the database;
Acquiring high-quality user duty ratio data in a second time period on a target media advertisement platform, wherein the duration of the second time period is longer than that of the first time period;
comparing and judging the high-quality user duty ratio in the first time period with the high-quality user duty ratio data in the second time period on a target media advertisement platform, and predicting the high-quality user duty ratio in a future third time period on the target media advertisement platform;
and adjusting the bid value of the advertisement position of the target media advertisement platform in real time according to the high-quality user duty ratio prediction result of the target media advertisement platform in a future third time period.
2. The method of bid adjustment of an ad slot of claim 1, further comprising, prior to the step of storing user quality information in the database:
acquiring user information of a user on a target media advertisement platform;
classifying the users according to the user information of the users, and grading each type of users;
summing the user scores corresponding to all the classifications of the user to obtain the quality score of the user;
acquiring quality scores of all users on a target media advertisement platform, arranging all the users according to a descending order of the quality scores, classifying the first twenty percent of users in the descending order as high-quality users, and classifying the last eighty percent of users in the descending order as low-quality users;
And forming the quality score information of the user by the quality score of the user and the quality classification result of the user.
3. The method for bid adjustment of an ad slot of claim 2, wherein the step of obtaining user information of a user on a target media advertisement platform during a first time period and calculating a high quality user duty cycle on the target media advertisement platform during the first time period in combination with the user quality information stored in the database comprises:
acquiring user information of a user in a first time period on a target media advertisement platform;
screening out the user information stored in the database from the user information of the user by combining the user quality information stored in the database;
acquiring corresponding quality scoring information according to the screened user information;
and calculating the high-quality user duty ratio of the target media advertisement platform in the first time period according to the acquired user quality scoring information.
4. The method of claim 3, wherein the step of calculating a high quality user duty cycle for a first period of time on the targeted media advertisement platform based on the obtained user quality score information comprises:
Obtaining a classification result of a user in the user quality scoring information;
the statistical classification result is the number A of the users with high quality, and the statistical classification result is the number B of the users with low quality;
and calculating the high-quality user duty ratio of the target media advertisement platform in the first time period according to the formula A/(A+B).
5. The bid adjustment method of advertisement space according to claim 2, wherein the classifying the using users according to the user information of the using users and scoring each type of users comprises:
acquiring user information of a user, wherein the user information comprises at least one of age, gender, occupation, preference, academic history, income, region, property, family condition, consumption level price, user click data and user watching duration;
constructing a classification model by using an XGBoost algorithm, wherein the classification model comprises a plurality of decision trees;
setting a predicted value of a leaf node at the tail end of each decision tree according to the classification parameters in each decision tree;
classifying the user information of the user into the terminal leaf nodes of the decision tree according to the user information of the user, and taking the predicted value of the terminal leaf nodes of the decision tree as the user score of the user corresponding to the decision tree.
6. The method of claim 1, wherein the XGBoost algorithm used constructs a classification model, the classification model comprising a plurality of decision tree steps, comprising:
setting a plurality of classification parameters in each decision tree, wherein the decision tree can classify a user to the end leaf node of the decision tree through the classification parameters;
is provided with
Representing a total of t blocksA value predicted by the strategy tree; />
Representing the predicted value of the 'previous t-1 decision tree' to the sample i; y is i Representing the actual value of sample i; f (f) t (x i ) Representing the predicted value of the t-th decision tree on the sample i; />
Representation and y i ，/>
A related loss function; omega (f) j ) Representing the model complexity of the j-th tree;
the predicted value of the classification model is set as follows:
the loss function of the classification model is set as follows:
wherein (1)>
Representation and y i ，/>
The loss function concerned, i.e. the deviation; />
A complex term representing the j-th tree;
simplifying the loss function:
wherein, the liquid crystal display device comprises a liquid crystal display device,
representing the first t-1 models;
taylor expansion of the loss function:
wherein g i f t (x i ) Is the first derivative of the residual error and,
is the second derivative of the residual;
the parameters of the tree are taken into the loss function:
wherein, the liquid crystal display device comprises a liquid crystal display device,
table omega j Showing the value of the jth leaf node in the tree; t represents the number of leaf nodes; i j Representing samples that are located at the j-th leaf node; />
G representing all samples belonging to the jth leaf node i Sum up; />
Represents h representing all samples belonging to the jth leaf node i Sum up; γt represents a quadratic function.
7. The method for bid adjustment of ad slots according to claim 1, wherein the step of adjusting the bid amount of the ad slots on the target media ad platform in real time based on the high quality user duty prediction result in the future third time period on the target media ad platform comprises:
establishing a bid number association relation between the high-quality user duty ratio and the advertisement position in a database;
acquiring bidding values of corresponding advertisement positions in a database according to a high-quality user duty ratio prediction result in a future third time period on the target media advertisement platform;
and feeding back the bid number of the corresponding advertisement position in the database to the target media advertisement platform as the bid of the advertisement position in a third time period in the future.
8. An apparatus for bid adjustment of an ad spot, the apparatus comprising:
the database management module is used for storing user quality information in the database, wherein the user quality information comprises user information and user quality scoring information;
The user quality detection module is used for acquiring user information of a user in a first time period on the target media advertisement platform and calculating the high-quality user duty ratio in the first time period on the target media advertisement platform by combining the user quality information stored in the database;
the historical data acquisition module is used for acquiring high-quality user duty ratio data in a second time period on the target media advertisement platform, wherein the duration of the second time period is longer than that of the first time period;
the user quality prediction module is used for comparing and judging the high-quality user duty ratio in the first time period with the high-quality user duty ratio data in the second time period on the target media advertisement platform and predicting the high-quality user duty ratio in a future third time period on the target media advertisement platform;
and the RTA strategy control bidding module is used for adjusting the bidding value of the advertising position of the target media advertising platform in real time according to the high-quality user duty ratio prediction result in the future third time period on the target media advertising platform.
9. A computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that the processor implements the steps of the method according to any one of claims 1 to 7 when the computer program is executed by the processor.
10. A computer readable storage medium, on which a computer program is stored, characterized in that the computer program, when being executed by a processor, implements the steps of the method of any of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310332940.XA CN116402558A (en) | 2023-03-31 | 2023-03-31 | Bid adjustment method, device, equipment and storage medium for advertisement space |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310332940.XA CN116402558A (en) | 2023-03-31 | 2023-03-31 | Bid adjustment method, device, equipment and storage medium for advertisement space |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116402558A true CN116402558A (en) | 2023-07-07 |
Family
ID=87009723
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310332940.XA Pending CN116402558A (en) | 2023-03-31 | 2023-03-31 | Bid adjustment method, device, equipment and storage medium for advertisement space |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116402558A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117217828A (en) * | 2023-08-18 | 2023-12-12 | 上海数禾信息科技有限公司 | Method, device, computer equipment and storage medium for verifying conversion return data |
-
2023
- 2023-03-31 CN CN202310332940.XA patent/CN116402558A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117217828A (en) * | 2023-08-18 | 2023-12-12 | 上海数禾信息科技有限公司 | Method, device, computer equipment and storage medium for verifying conversion return data |
| CN117217828B (en) * | 2023-08-18 | 2024-05-10 | 上海数禾信息科技有限公司 | Method, device, computer equipment and storage medium for verifying conversion return data |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US10607252B2 (en) | Methods and systems for targeted B2B advertising campaigns generation using an AI recommendation engine | |
| US11176495B1 (en) | Machine learning model ensemble for computing likelihood of an entity failing to meet a target parameter | |
| US20200342489A1 (en) | Methods and systems for b2b demand generation with targeted advertising campaigns and lead profile optimization based on target audience feedback | |
| CN109783730A (en) | Products Show method, apparatus, computer equipment and storage medium | |
| US11727328B2 (en) | Machine learning systems and methods for predictive engagement | |
| US20180375947A1 (en) | Automated communication-based intelligence engine | |
| US20220398610A1 (en) | Method of forecasting store demand based on artificial intelligence and system therefor | |
| US20190057404A1 (en) | Jobs forecasting | |
| US20230230183A1 (en) | Intelligent Prediction of An Expected Value of User Conversion | |
| CN110135943B (en) | Product recommendation method, device, computer equipment and storage medium | |
| CN112633962A (en) | Service recommendation method and device, computer equipment and storage medium | |
| CN113627566A (en) | Early warning method and device for phishing and computer equipment | |
| CN116402558A (en) | Bid adjustment method, device, equipment and storage medium for advertisement space | |
| CN115423578B (en) | Bid bidding method and system based on micro-service containerized cloud platform | |
| CN115696533A (en) | Communication base station energy consumption optimization method and equipment based on 5G | |
| US20200302400A1 (en) | Job-post budget recommendation based on performance | |
| CN108681871A (en) | A kind of method of prompt message, terminal device and computer readable storage medium | |
| CN117196322B (en) | Intelligent wind control method, intelligent wind control device, computer equipment and storage medium | |
| CN112685639A (en) | Activity recommendation method and device, computer equipment and storage medium | |
| CN115587844A (en) | Method, device, equipment and storage medium for controlling user conversion data return | |
| CN116912016A (en) | Bill auditing method and device | |
| US11210685B2 (en) | System and method for multi-channel digital marketing optimization | |
| CN117217828B (en) | Method, device, computer equipment and storage medium for verifying conversion return data | |
| Heinrich et al. | A quantitative approach for modelling the influence of currency of information on decision-making under uncertainty | |
| CN117689427A (en) | User bidding coefficient control method, device, computer equipment and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |