Advertisement recall method and device, storage medium and electronic equipment

Abstract

The invention relates to an advertisement recall method and device, a storage medium and electronic equipment. The method comprises the following steps: calculating the historical display cost and price adjustment sensitivity coefficient of each advertisement in each traffic scene according to the real-time bid of each advertisement in each traffic scene; calculating a recall score of each advertisement in each traffic scene according to the historical display cost and the price adjustment sensitivity coefficient; sorting all the advertisements according to the recall scores of all the advertisements in each traffic scene to obtain recalled advertisement sorting data of each traffic scene; and under the condition that it is detected that the target traffic scene has the traffic request, the previous target number of advertisements in the recalled advertisement sorting data of the target traffic scene are taken to enter a recall link, and the target traffic scene is any one traffic scene in all the traffic scenes. The technical problem that the advertisement sorting is inconsistent and the putting effect is affected due to the fact that the recall link does not respond to real-time bid adjustment in time is solved.

Description

Technical Field

[0001] This application relates to the field of computer technology, and in particular, to a method, device, storage medium, and electronic device for ad recall. Background Art

[0002] In the ad delivery pipeline, recall, rough ranking, fine ranking, and re-ranking are four key steps. As the most front-end part of the delivery pipeline, the recall step directly determines which ads can enter the candidate pool. Therefore, it is crucial for the ad delivery volume. Usually, customers adjust the bid price according to their buying volume target and real-time running volume to control the ad display volume and ensure that the ad exposure volume can reach the expected target. However, with the increase in the types and quantities of ads on the ad delivery platform, how to effectively screen out suitable ads from numerous candidate ads has become a challenge. To ensure the effectiveness of ad delivery, the vast majority of ad platforms adopt a ranking mechanism based on the historical CPM (Cost Per Mille) of ads in the recall step. CPM is an important indicator to measure the ad display effect and cost. Its calculation method is to divide the ad cost by the number of exposures and then multiply by 1000, aiming to reflect the cost required for every thousand displays. Ranking ads by historical CPM can preferentially select ads with better performance to enter the recall candidate pool, thereby improving the efficiency and quality of delivery. Usually, the platform will select the top few ranked ads to enter the subsequent delivery steps. However, there is a significant problem with the existing recall mechanism: the real-time bid information of ads is not used in the recall step. Since customers will adjust the ad bid strategy in real time to meet different delivery requirements, this adjustment usually does not immediately reflect in the recall step. Therefore, the ad ranking in the recall stage may be inconsistent with the subsequent ranking steps, resulting in high-quality ads not being recalled in time, thus affecting their display opportunities and delivery effects. This inconsistency will lead to a decrease in the effectiveness and efficiency of ad delivery. Especially in scenarios with fierce competition or frequent bid adjustments, advertisers may not be able to optimize their ad displays through timely price adjustments, thereby affecting the overall delivery effect of the platform. Summary of the Invention

[0003] This application provides a method, device, storage medium, and electronic device for ad recall to solve the technical problem that the recall step fails to respond in time to real-time bid adjustments, resulting in inconsistent ad rankings and affecting the delivery effect.

[0004] In a first aspect, the present application provides a method for recalling advertisements, including: calculating the historical display cost and price adjustment sensitivity coefficient of each of the above advertisements in each of the above traffic scenarios according to the real-time bid of each advertisement in each traffic scenario; calculating the recall score of each of the above advertisements in each of the above traffic scenarios according to the above historical display cost and the above price adjustment sensitivity coefficient; sorting all the above advertisements according to the recall scores of all the above advertisements in each of the above traffic scenarios to obtain the recall advertisement sorting data of each of the above traffic scenarios; and when a traffic request is detected in a target traffic scenario, taking the top target number of advertisements in the recall advertisement sorting data of the above target traffic scenario into the recall link, where the above target traffic scenario is any one of all the above traffic scenarios.

[0005] In a second aspect, the present application provides an advertisement recall device, including: a first calculation module for calculating the historical display cost and price adjustment sensitivity coefficient of each of the above advertisements in each of the above traffic scenarios according to the real-time bid of each advertisement in each traffic scenario; a second calculation module for calculating the recall score of each of the above advertisements in each of the above traffic scenarios according to the above historical display cost and the above price adjustment sensitivity coefficient; a sorting module for sorting all the above advertisements according to the recall scores of all the above advertisements in each of the above traffic scenarios to obtain the recall advertisement sorting data of each of the above traffic scenarios; and a recall module for taking the top target number of advertisements in the recall advertisement sorting data of the above target traffic scenario into the recall link when a traffic request is detected in the target traffic scenario, where the above target traffic scenario is any one of all the above traffic scenarios.

[0006] As an optional example, the above first calculation module includes: a first acquisition unit for acquiring the historical revenue and historical exposure times of each of the above advertisements in each of the above traffic scenarios; a first calculation unit for calculating the historical display cost of each of the above advertisements in each of the above traffic scenarios through the following formula: where the above CPM i,j is the historical display cost of the i-th advertisement among all the above advertisements in the j-th traffic scenario among all the above traffic scenarios, the above In i,j is the historical revenue of the i-th advertisement in the j-th traffic scenario, and the above Im i,j is the historical exposure times of the i-th advertisement in the j-th traffic scenario.

[0007] As an optional example, the above-mentioned first calculation unit includes: an acquisition subunit, configured to, in the case that there is no historical revenue and historical exposure times for the first advertisement, acquire the historical display costs of all advertisements in the first traffic scenario, where the first advertisement is any one of all the advertisements, and the first traffic scenario is each traffic scenario among all the traffic scenarios; a calculation subunit, configured to calculate the average historical display cost in the first traffic scenario according to the historical display costs of all advertisements in the first traffic scenario; and a determination subunit, configured to determine the average historical display cost as the historical display cost of the first advertisement in the first traffic scenario.

[0008] As an optional example, the above-mentioned first calculation module includes: a second acquisition module, configured to acquire the historical average bid and real-time bid of each of the above-mentioned advertisements in each of the above-mentioned traffic scenarios; and a second calculation unit, configured to calculate the price adjustment sensitivity coefficient of each of the above-mentioned advertisements in each of the above-mentioned traffic scenarios through the following formula: where the above-mentioned K i,j is the price adjustment sensitivity coefficient of the i-th advertisement among all the above-mentioned advertisements in the j-th traffic scenario among all the above-mentioned traffic scenarios, the above-mentioned bid_new i,j is the real-time bid of the i-th advertisement in the j-th traffic scenario, and the above-mentioned bid_old i,j is the historical average bid of the i-th advertisement in the j-th traffic scenario.

[0009] As an optional example, the above-mentioned second calculation module includes: a third calculation unit, configured to calculate the recall score of each of the above-mentioned advertisements in each of the above-mentioned traffic scenarios through the following formula: S i,j = CPM i,j * K i,j i = 1, 2... m, j = 1, 2... n; where the above-mentioned S i,j is the recall score of the i-th advertisement among all the above-mentioned advertisements in the j-th traffic scenario among all the above-mentioned traffic scenarios, the above-mentioned CPM i,j is the historical display cost of the i-th advertisement in the j-th traffic scenario, and the above-mentioned K i,j is the price adjustment sensitivity coefficient of the i-th advertisement in the j-th traffic scenario.

[0010] As an optional example, the above-mentioned sorting module includes: a processing unit, configured to use each of the above-mentioned traffic scenarios as the current traffic scenario, and perform the following operations on the current traffic scenario: acquire the recall scores of all the above-mentioned advertisements in the current traffic scenario; and sort all the above-mentioned advertisements in descending order according to the recall scores to obtain the recall advertisement sorting data of the current traffic scenario.

[0011] As an alternative example, after obtaining the recall advertisement sorting data in each of the above traffic scenarios, save each of the above recall advertisement sorting data to a target database. The above recall module includes: a third acquisition module, configured to acquire the recall advertisement sorting data of the target traffic scenario from the above target database; a recall unit, configured to select the top target number of advertisements in the recall advertisement sorting data of the above target traffic scenario to enter the recall process.

[0012] In a third aspect, the present application provides a storage medium, in which a computer program is stored. Wherein, when the computer program is run by a processor, it executes the above-mentioned advertisement recall method.

[0013] In a fourth aspect, the present application further provides an electronic device, including a memory and a processor. A computer program is stored in the above memory, and the above processor is configured to execute the above-mentioned advertisement recall method through the computer program.

[0014] The above technical solution provided by the embodiments of the present application has the following advantages compared with the prior art:

[0015] The present application calculates the historical display cost and price adjustment sensitivity coefficient of each advertisement in each traffic scenario according to the real-time bid of each advertisement in each traffic scenario; calculates the recall score of each advertisement in each traffic scenario according to the above historical display cost and the above price adjustment sensitivity coefficient; sorts all the above advertisements according to the recall scores of all the above advertisements in each traffic scenario to obtain the recall advertisement sorting data of each traffic scenario; when a traffic request for a target traffic scenario is detected, select the top target number of advertisements in the recall advertisement sorting data of the above target traffic scenario to enter the recall process, where the above target traffic scenario is any one of all the above traffic scenarios. Since in the above method, based on the real-time bid of each advertisement in different traffic scenarios, its historical display cost and price adjustment sensitivity coefficient are calculated, and then, according to the historical display cost and price adjustment sensitivity coefficient, the recall score of each advertisement is calculated, and by sorting the recall scores of the advertisements, the recall advertisement sorting data of each traffic scenario is obtained. When a traffic request occurs in a certain traffic scenario, the recall advertisement sorting data of this scenario is read in real time, and the top target number of advertisements is selected for recall and delivery, thereby achieving an accurate match between advertisements and traffic scenarios, responding to bid adjustments, reducing the lag in the recall and delivery links, improving the accuracy and efficiency of advertisement delivery, improving the quality and effect of advertisement display, and ensuring that advertisement delivery is more effective, and further solving the technical problem that the recall link fails to respond to real-time bid adjustments in a timely manner, resulting in inconsistent advertisement sorting and affecting the delivery effect. Description of the Drawings

[0016] The accompanying drawings herein are incorporated into and constitute a part of this specification, showing embodiments consistent with the present application and, together with the specification, are used to explain the principles of the present application.

[0017] To more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the accompanying drawings required for use in the description of the embodiments or the prior art. Obviously, for those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0018] One or more embodiments are exemplarily illustrated by the pictures in the corresponding accompanying drawings. These exemplary illustrations do not constitute limitations on the embodiments. Elements with the same reference numerals in the drawings represent similar elements, unless otherwise stated, and the drawings in the figures do not constitute a proportional limitation.

[0019] Figure 1 is a flowchart of an optional advertisement recall method according to an embodiment of the present application;

[0020] Figure 2 is a block diagram of the implementation of an optional advertisement recall method according to an embodiment of the present application;

[0021] Figure 3 is a schematic structural diagram of an optional advertisement recall device according to an embodiment of the present application;

[0022] Figure 4 is a schematic diagram of an optional electronic device according to an embodiment of the present application. Detailed implementation manners

[0023] To make the objectives, technical solutions, and advantages of the embodiments of the present application clearer, the following will clearly and completely describe the technical solutions in the embodiments of the present application with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are some but not all of the embodiments of the present application. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present application without creative efforts fall within the scope of protection of the present application.

[0024] The following disclosure provides many different embodiments or examples for implementing different structures of the present application. To simplify the disclosure of the present application, the components and settings of specific examples are described below. Of course, they are only examples and are not intended to limit the present application. In addition, the present application may repeat reference numerals and / or letters in different examples. This repetition is for the purpose of simplification and clarity and does not itself indicate the relationship between the various embodiments and / or settings discussed.

[0025] According to the first aspect of the embodiments of the present application, a method for recalling advertisements is provided. Optionally, as Figure 1 shown, the above method includes:

[0026] S102, calculate the historical display cost and price adjustment sensitivity coefficient of each advertisement in each traffic scenario according to the real-time bid of each advertisement in each traffic scenario;

[0027] S104, calculate the recall score of each advertisement in each traffic scenario according to the historical display cost and price adjustment sensitivity coefficient;

[0028] S106, sort all advertisements according to the recall scores of all advertisements in each traffic scenario to obtain the recall advertisement sorting data for each traffic scenario;

[0029] S108, in the case of detecting a traffic request for a target traffic scenario, select the top target number of advertisements in the recall advertisement sorting data of the target traffic scenario to enter the recall link, where the target traffic scenario is any one of all traffic scenarios.

[0030] Optionally, in this embodiment, the real-time bid of each advertisement in different traffic scenarios changes dynamically, and the advertising effect of each traffic scenario will also fluctuate accordingly. To measure the performance of advertisements in each traffic scenario, it is first necessary to calculate the historical display cost CPM. CPM (cost per mille) refers to the fee that an advertiser needs to pay to have an advertisement displayed 1000 times in a specified traffic scenario. The historical display cost CPM reflects the advertising placement effect in a specific traffic scenario. In addition, it is also necessary to calculate the price adjustment sensitivity coefficient of each advertisement in each traffic scenario according to the real-time bid, that is, the sensitivity of the advertisement's response effect when adjusting the bid. This coefficient reflects the impact of the advertisement bid change on the placement effect and ensures the smoothness of the placement effect after the bid adjustment. According to the historical display cost and price adjustment sensitivity coefficient, calculate the recall score of each advertisement in each traffic scenario. The recall score comprehensively considers the advertising placement cost and price adjustment response ability and is used to measure the comprehensive performance of the advertisement in this traffic scenario. The higher the recall score, the better the performance of the advertisement in this traffic scenario and the more suitable it is to enter the recall candidate pool. In each traffic scenario, all advertisements are sorted according to their recall scores. After sorting, select the top multiple advertisements to enter the recall candidate pool for further placement. This sorting process can ensure that the platform displays the best-performing advertisements in each traffic scenario. When detecting a traffic request for a certain target traffic scenario, according to the recall advertisement sorting data of this scenario, select the top multiple advertisements to enter the recall link for subsequent advertisement bidding and placement. The target traffic scenario refers to any specific scenario among all traffic scenarios of the platform, and the target number is the predetermined number of advertisement placements.

[0031] Optionally, in this embodiment, by combining the historical display cost and the price adjustment sensitivity coefficient, the performance of the advertisement in each traffic scenario can be evaluated more accurately, so as to ensure that the most suitable advertisement is selected in the recall process. The introduction of the price adjustment sensitivity coefficient enables the system to better respond to the bid adjustment of the advertiser, reduces the recall inconsistency caused by the bid change, and improves the flexibility and efficiency of the advertisement placement. Sorting the advertisements according to the recall score can ensure that efficient and high-quality advertisements enter the candidate pool, improve the accuracy of the advertisement placement, and reduce the display of invalid advertisements. Through the refined recall and sorting mechanisms, the effect of the advertisement placement is optimized, the bid adjustment of the advertiser and the placement strategy of the platform are more matched, thereby improving the overall placement effect and efficiency of the advertisement platform.

[0032] As an optional example, calculating the historical display cost and the price adjustment sensitivity coefficient of each advertisement in each traffic scenario according to the real-time bid of each advertisement in each traffic scenario includes:

[0033] Obtain the historical revenue and the historical exposure times of each advertisement in each traffic scenario;

[0034] Calculate the historical display cost of each advertisement in each traffic scenario through the following formula:

[0035]

[0036] where, CPM i,j is the historical display cost of the i-th advertisement among all advertisements in the j-th traffic scenario among all traffic scenarios, In i,j is the historical revenue of the i-th advertisement in the j-th traffic scenario, Im i,j is the historical exposure times of the i-th advertisement in the j-th traffic scenario.

[0037] Optionally, in this embodiment, during the advertisement placement process, the historical display cost and the price adjustment sensitivity coefficient are two important measurement indicators for evaluating the performance of the advertisement in different traffic scenarios and its response to the bid adjustment. For each advertisement in each traffic scenario, it is necessary to collect its historical revenue and historical exposure times in the past period. The historical revenue refers to the total revenue obtained by the advertisement in a specific traffic scenario, and the historical exposure times indicate the number of times the advertisement is displayed in this scenario. Through these two data, the effect of the advertisement in this traffic scenario can be measured. According to the historical revenue and historical exposure times of the advertisement, use the following formula to calculate the historical display cost of each advertisement in each traffic scenario:

[0038]

[0039] where, CPM i,jis the historical display cost of the i-th advertisement in the j-th traffic scenario, In i,j is the historical revenue of the ith advertisement in the jth traffic scenario, Im i,j is the historical exposure count of the ith advertisement in the jth traffic scenario. Through this formula, the historical display cost of the advertisement in each traffic scenario can be obtained, reflecting the cost and effect of the advertisement.

[0040] Through historical display cost calculation, we can clearly understand the effect and cost of advertising in different traffic scenarios, helping advertisers achieve more accurate cost management. Using historical revenue and exposure data to support advertising decisions can make more reasonable advertising plans based on data, reduce invalid advertising displays, and improve the return on investment of advertising.

[0041] As an optional example, calculating the historical display cost of each ad in each traffic scenario includes:

[0042] In the case where the first advertisement has no historical revenue and historical exposure times, the historical display costs of all advertisements in the first traffic scenario are obtained, wherein the first advertisement is any one of all advertisements, and the first traffic scenario is each traffic scenario in all traffic scenarios;

[0043] According to the historical display costs of all advertisements in the first traffic scenario, the average historical display cost in the first traffic scenario is calculated;

[0044] The average historical display cost is determined as the historical display cost of the first advertisement in the first traffic scenario.

[0045] Optionally, in this embodiment, in some cases, the first advertisement (i.e., a specific advertisement) is a new advertisement and has no historical revenue and historical exposure times. In this case, it is necessary to estimate its historical display cost based on other advertisements in the traffic scenario where the advertisement is located. First, it is necessary to collect the historical display costs of all advertisements in the first traffic scenario (i.e., any one of all traffic scenarios). After collecting the historical display costs of all advertisements, the average historical display cost of all advertisements in the first traffic scenario is calculated next (i.e., the average historical display cost of all advertisements in the first traffic scenario). For the first advertisement (i.e., an advertisement that has no historical revenue and historical exposure times), the average historical display cost of all advertisements in the first traffic scenario is used as the historical display cost of the advertisement in the traffic scenario.

[0046] By using the historical display costs of other ads in the same traffic scenario to infer the display costs of new ads, the gap of new ads having no historical revenue and exposure times is effectively filled, avoiding calculation interruptions caused by missing data, thereby ensuring smooth and consistent calculation of historical display costs of ads.

[0047] As an optional example, calculating the historical display cost and price adjustment sensitivity coefficient of each advertisement in each traffic scenario according to the real-time bid of each advertisement in each traffic scenario includes:

[0048] Obtain the historical average bid and real-time bid of each advertisement in each traffic scenario;

[0049] Calculate the price adjustment sensitivity coefficient of each advertisement in each traffic scenario through the following formula:

[0050]

[0051] where K i,j is the price adjustment sensitivity coefficient of the i-th advertisement among all advertisements in the j-th traffic scenario among all traffic scenarios, bid_new i,j is the real-time bid of the i-th advertisement in the j-th traffic scenario, bid_old i,j is the historical average bid of the i-th advertisement in the j-th traffic scenario.

[0052] Optionally, in this embodiment, according to the bid data of the advertisement in different traffic scenarios, the price adjustment sensitivity coefficient is used to measure the impact of the advertisement's bid adjustment on the display cost. This coefficient helps to evaluate the competitiveness and display possibility of the advertisement in this traffic scenario after the advertisement's bid adjustment. First, for each advertisement in each traffic scenario, two bid data need to be obtained: Historical average bid: The historical average bid of this advertisement in this traffic scenario, which reflects the bid level of the advertisement in the past in this scenario; Real-time bid: The current bid of this advertisement in this traffic scenario, which reflects the real-time bid adjustment of the advertiser. Use the following formula to calculate the price adjustment sensitivity coefficient of each advertisement in each traffic scenario:

[0053]

[0054] where K i,j is the price adjustment sensitivity coefficient of the i-th advertisement in the j-th traffic scenario, bid_new i,j is the real-time bid of the i-th advertisement in the j-th traffic scenario, bid_old i,j is the historical average bid of the i-th advertisement in the j-th traffic scenario.

[0055] If bid_new i,j > bid_old i,j , that is, the real-time bid of the advertisement is higher than the historical bid, then the price adjustment sensitivity coefficient K i,j will be greater than 1, indicating that the advertisement is more sensitive to bid adjustment and may increase the opportunity for advertisement display. If bid_new i,j= bid_old i,j , then the price adjustment sensitivity coefficient K i,j will be equal to 1, indicating that the bid has not changed and the performance of the advertisement in this traffic scenario has not changed. If bid_new i,j < bid_old i,j , then the price adjustment sensitivity coefficient K i,j is less than 1, indicating that the advertisement is not very sensitive to the bid reduction, which may lead to a reduction in the display opportunities of the advertisement. To avoid the influence of extreme values, upper and lower limits are usually set for the price adjustment sensitivity coefficient. The maximum value is 10, that is, the advertisement responds very strongly to the bid, and the minimum value is 0.1, that is, the advertisement responds very weakly to the bid. These upper and lower limits can ensure the smoothness of the price adjustment sensitivity coefficient and avoid system instability caused by data fluctuations.

[0056] The price adjustment sensitivity coefficient can quantify the response of the advertisement to the bid adjustment, help the platform understand the impact of the advertisement bid change on the display opportunities, and optimize the advertisement placement strategy. Through the price adjustment sensitivity coefficient, the advertiser can adjust the bid strategy according to the performance of the advertisement to obtain better display effects. The price adjustment sensitivity coefficient can reflect the advertisement placement situation in real time and ensure the consistency of advertisement display and bid adjustment. In the advertisement placement system, the price adjustment sensitivity coefficient can help the system more accurately evaluate the placement potential of the advertisement and effectively sort the advertisements, thereby improving the efficiency and quality of advertisement placement. By setting the upper and lower limits of the price adjustment sensitivity coefficient, system instability or uneven advertisement display caused by extreme data is avoided, thereby improving the placement effect and stability of the platform.

[0057] As an optional example, calculating the recall score of each advertisement in each traffic scenario according to the historical display cost and the price adjustment sensitivity coefficient includes:

[0058] Calculate the recall score of each advertisement in each traffic scenario through the following formula:

[0059] S i,j = CPM i,j * K i,j i = 1, 2... m, j = 1, 2... n;

[0060] where, S i,j is the recall score of the i-th advertisement among all advertisements in the j-th traffic scenario among all traffic scenarios, CPM i,j is the historical display cost of the i-th advertisement in the j-th traffic scenario, and K i,j is the price adjustment sensitivity coefficient of the i-th advertisement in the j-th traffic scenario.

[0061] Optionally, in this embodiment, the recall score S i,jUsed to evaluate the comprehensive performance of each advertisement in each traffic scenario, which combines the historical display cost CPM of the advertisement i,j and the price adjustment sensitivity coefficient K i,j , so as to reflect the comprehensive value of the advertisement and the sensitivity of bid adjustment. First, calculate its historical display cost according to the historical revenue and historical exposure times of the advertisement, then calculate the price adjustment sensitivity coefficient according to the real-time bid and the historical average bid, and finally multiply the historical display cost by the price adjustment sensitivity coefficient to obtain the recall score.

[0062] A high recall score indicates that the advertisement has a high display cost in the current traffic scenario and is sensitive to bid adjustment, and may be more in line with the placement requirements. A low recall score indicates that the advertisement has a low display cost in the current traffic scenario, or the bid adjustment has little impact on it, and it may have a lower priority. Through the recall score, advertisements can be sorted in each traffic scenario, and advertisements with higher recall scores are preferentially selected to enter the subsequent links, thereby improving the placement efficiency and effect of advertisements. The recall score comprehensively considers the historical display cost and the price adjustment sensitivity coefficient, can timely reflect the performance of the advertisement in the current traffic scenario, and is convenient for the advertisement to actively adjust the bid strategy. Through the calculation of the recall score, advertisements with better performance can be preferentially screened out, thereby improving the accuracy and overall revenue of advertisement placement. The calculation of the recall score takes into account both historical data (display cost) and real-time data (price adjustment sensitivity coefficient), realizing the comprehensive optimization of advertisement ranking.

[0063] As an optional example, according to the recall scores of all advertisements in each traffic scenario, all advertisements are sorted, and the recall advertisement sorting data for each traffic scenario includes:

[0064] Take each traffic scenario as the current traffic scenario, and perform the following operations on the current traffic scenario:

[0065] Obtain the recall scores of all advertisements in the current traffic scenario;

[0066] Sort all advertisements in descending order of the recall score to obtain the recall advertisement sorting data for the current traffic scenario.

[0067] Optionally, in this embodiment, recall ad ranking refers to sorting ads in descending order of recall scores in a specific traffic scenario to generate an ordered ad list, which is used to guide subsequent ad screening and display strategies. Specifically, each traffic scenario is used as the currently processed traffic scenario, and the sorting operation is performed one by one to obtain the recall scores of all ads in the current traffic scenario, which have been calculated through the previous formula. The ads are sorted according to the recall scores, from largest to smallest, and the sorted ad list is used as the recall ad ranking data for the current traffic scenario and saved. For each traffic scenario, the recall ad ranking data ensures that high-priority ads can enter the subsequent links first, improving the overall effect of ad delivery. The ranking data for each traffic scenario is generated independently, enabling flexible adjustment of ad priorities for different traffic environments and enhancing scenario adaptation capabilities. By sorting the recall scores, high-value ads can be intuitively screened out to ensure that these ads can enter the delivery link first.

[0068] As an optional example, after obtaining the recall ad ranking data for each traffic scenario, each recall ad ranking data is saved to the target database. When a traffic request for the target traffic scenario is detected, taking the top target number of ads from the recall ad ranking data of the target traffic scenario into the recall link includes:

[0069] Obtain the recall ad ranking data of the target traffic scenario from the target database;

[0070] Take the top target number of ads from the recall ad ranking data of the target traffic scenario into the recall link.

[0071] Optionally, in this embodiment, the target database can be a Couchbase database. Couchbase is a high-performance, distributed NoSQL database that combines the advantages of relational databases and traditional NoSQL databases and is particularly suitable for real-time, high-concurrency data storage and query scenarios.

[0072] Optionally, in this embodiment, after sorting the recalled advertisements for each traffic scenario, the sorted data needs to be stored in a target database so as to quickly respond when subsequent traffic requests arrive. The stored sorted data may include, but is not limited to, the following: the identifier of the traffic scenario (such as scenario ID or name), the list of advertisement IDs and their sorting orders, the generation time of the sorted data, etc. When it is detected that a certain target traffic scenario issues a traffic request, operations need to be performed on the specific sorted data for that scenario. The target traffic scenario is defined as a specific traffic scenario with current placement requirements. Retrieve the corresponding recalled advertisement sorted data from the target database according to the identifier of the traffic scenario, and load the recalled advertisement sorted data of the target traffic scenario into the system for subsequent use. The target quantity is a predefined integer value (for example, the top 100 advertisements), indicating the number of advertisements to be selected from the sorted data. Take the top target quantity of advertisements from the recalled advertisement sorted data in order to form the advertisement list entering the recall link. The selected advertisement list is directly passed to the recall link for subsequent advertisement screening, sorting, and placement link usage.

[0073] By pre-storing the sorted data, it can be quickly retrieved and processed when a traffic request occurs, improving the real-time response ability of the system. Intercepting high-priority advertisements according to the target quantity avoids resource consumption for processing all advertisements and improves the system efficiency. According to the sorting score and the target quantity, it is ensured that the advertisements entering the recall link have high placement value, optimizing the advertisement effect.

[0074] An example is combined for illustration. This application relates to a method for recalling advertisements. To improve the consistency of the effective advertisement placement link and reduce the loss of high-quality advertisements in each link, in the recall link of advertisement placement, support for the sensitivity of advertisement price adjustment is provided. Combining the real-time bid of the advertisement and the average bid within a certain historical period, a smoothing formula is designed to calculate the price adjustment sensitivity coefficient of the advertisement. When calculating the recall sorting of the advertisement in the recall link, multiply the price adjustment sensitivity coefficient of the advertisement by the historical CPM of the advertisement, and reflect the advertisement price adjustment information in the final recall advertisement sorting in real time, which can effectively improve the consistency of the advertisement placement link, ensure that advertisements with real-time increased bids have a higher probability of entering the recall link, and reduce the probability of advertisements with real-time price cuts entering the recall link. Thus, the purpose of allowing customers to control the running volume speed by adjusting the bid is achieved, as well as the goal of maximizing the platform's advertisement placement revenue. Specifically, the implementation block diagram is as Figure 2 shown:

[0075] 1. Calculate the historical display cost: Count the revenue and exposure times of each advertisement in different traffic scenarios in the past 7 days, and calculate the historical display cost of the advertisement through the formula historical display cost CPM = revenue / exposure count * 1000. If the advertisement is a new advertisement, take the average CPM of all advertisements in that traffic scenario for initialization;

[0076] 2. Calculate the price adjustment sensitivity coefficient: First, calculate the average bid of the advertisement in the past 7 days, denoted as bid_old, and read the real-time bid of the advertisement, denoted as bid_new. To ensure the smoothness of the price adjustment sensitivity coefficient, design a mathematical formula: price adjustment sensitivity coefficient = 1 + log(bid_new / bid_old) to calculate the price adjustment sensitivity coefficient. In addition, to avoid the instability of advertisement delivery caused by factors such as customers' misoperations, design the upper and lower threshold values of the price adjustment sensitivity coefficient, with the maximum value being 10 and the minimum value being 0.1;

[0077] 3. Calculate the recall score: Multiply the historical display cost of the advertisement by the price adjustment sensitivity coefficient to calculate the recall score of the advertisement;

[0078] 4. Recall advertisement sorting data: For each traffic scenario, sort all advertisements from large to small based on the recall score to obtain the recall advertisement sorting data for each traffic scenario, and then store all the recall advertisement sorting data in the Couchbase database;

[0079] 5. Advertisement recall: When there is a traffic request for a certain traffic scenario, read the corresponding recall advertisement sorting data from the CouchBase database in real time, select the top several advertisements for recall, and then go through the subsequent sorting process for competitive delivery.

[0080] It should be noted that for the foregoing method embodiments, for the sake of simple description, they are all expressed as a series of action combinations. However, those skilled in the art should know that this application is not limited by the described action sequence, because according to this application, certain steps can be performed in other sequences or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily essential to this application.

[0081] On the other hand, according to an embodiment of the present application, there is also provided a device for recalling advertisements, as Figure 3 shown, including:

[0082] The first calculation module 302 is used to calculate the historical display cost and the price adjustment sensitivity coefficient of each advertisement in each traffic scenario according to the real-time bid of each advertisement in each traffic scenario;

[0083] The second calculation module 304 is used to calculate the recall score of each advertisement in each traffic scenario according to the historical display cost and the price adjustment sensitivity coefficient;

[0084] The sorting module 306 is used to sort all advertisements according to the recall scores of all advertisements in each traffic scenario to obtain the recall advertisement sorting data for each traffic scenario;

[0085] A recall module 308, configured to, when a traffic request is detected in a target traffic scenario, select the top target number of advertisements from the recall advertisement sorting data of the target traffic scenario to enter the recall process, where the target traffic scenario is any one of all traffic scenarios.

[0086] It should be noted that the first calculation module 302 in this embodiment may be used to execute step S102 in the embodiment of the present application, the second calculation module 304 in this embodiment may be used to execute step S104 in the embodiment of the present application, the sorting module 306 in this embodiment may be used to execute step S106 in the embodiment of the present application, and the recall module 308 in this embodiment may be used to execute step S108 in the embodiment of the present application.

[0087] As an alternative example, the first calculation module includes:

[0088] A first acquisition unit, configured to acquire the historical revenue and historical exposure times of each advertisement in each traffic scenario;

[0089] A first calculation unit, configured to calculate the historical display cost of each advertisement in each traffic scenario through the following formula:

[0090]

[0091] where CPM i,j is the historical display cost of the i-th advertisement among all advertisements in the j-th traffic scenario among all traffic scenarios, In i,j is the historical revenue of the i-th advertisement in the j-th traffic scenario, and Im i,j is the historical exposure times of the i-th advertisement in the j-th traffic scenario.

[0092] As an alternative example, the first calculation unit includes:

[0093] An acquisition subunit, configured to, when the first advertisement has no historical revenue and historical exposure times, acquire the historical display costs of all advertisements in the first traffic scenario, where the first advertisement is any one of all advertisements, and the first traffic scenario is each traffic scenario among all traffic scenarios;

[0094] A calculation subunit, configured to calculate the average historical display cost in the first traffic scenario according to the historical display costs of all advertisements in the first traffic scenario;

[0095] A determination subunit, configured to determine the average historical display cost as the historical display cost of the first advertisement in the first traffic scenario.

[0096] As an alternative example, the first calculation module includes:

[0097] A second acquisition module, configured to acquire the historical average bid and the real-time bid of each advertisement in each traffic scenario;

[0098] A second calculation unit, configured to calculate the price adjustment sensitivity coefficient of each advertisement in each traffic scenario through the following formula:

[0099]

[0100] where K i,j is the price adjustment sensitivity coefficient of the i-th advertisement among all advertisements in the j-th traffic scenario among all traffic scenarios, bid_new i,j is the real-time bid of the i-th advertisement in the j-th traffic scenario, and bid_old i,j is the historical average bid of the i-th advertisement in the j-th traffic scenario.

[0101] As an optional example, the second calculation module includes:

[0102] A third calculation unit, configured to calculate the recall score of each advertisement in each traffic scenario through the following formula:

[0103] S i,j = CPM i,j * K i,j i = 1, 2... m, j = 1, 2... n;

[0104] where S i,j is the recall score of the i-th advertisement among all advertisements in the j-th traffic scenario among all traffic scenarios, CPM i,j is the historical display cost of the i-th advertisement in the j-th traffic scenario, and K i,j is the price adjustment sensitivity coefficient of the i-th advertisement in the j-th traffic scenario.

[0105] As an optional example, the sorting module includes:

[0106] A processing unit, configured to use each traffic scenario as the current traffic scenario, and perform the following operations on the current traffic scenario:

[0107] Acquire the recall scores of all advertisements in the current traffic scenario;

[0108] Sort all advertisements in descending order of the recall scores to obtain the recall advertisement sorting data of the current traffic scenario.

[0109] As an optional example, after obtaining the recall advertisement sorting data of each traffic scenario, save each recall advertisement sorting data to a target database. The recall module includes:

[0110] A third acquisition module, configured to acquire the recall advertisement ranking data of the target traffic scenario from the target database;

[0111] A recall unit, configured to select the top target number of advertisements from the recall advertisement ranking data of the target traffic scenario to enter the recall link.

[0112] For other examples of this embodiment, please refer to the above examples and will not be elaborated here.

[0113] Figure 4 It is a schematic diagram of an optional electronic device according to an embodiment of the present application. As Figure 4 shown, it includes a processor 402, a communication interface 404, a memory 406, and a communication bus 408. Among them, the processor 402, the communication interface 404, and the memory 406 complete mutual communication through the communication bus 408. Among them,

[0114] The memory 406 is configured to store a computer program;

[0115] The processor 402, when executing the computer program stored on the memory 406, implements the following steps:

[0116] Calculate the historical display cost and price adjustment sensitivity coefficient of each advertisement in each traffic scenario according to the real-time bid price of each advertisement in each traffic scenario;

[0117] Calculate the recall score of each advertisement in each traffic scenario according to the historical display cost and price adjustment sensitivity coefficient;

[0118] Sort all advertisements according to the recall scores of all advertisements in each traffic scenario to obtain the recall advertisement ranking data of each traffic scenario;

[0119] When it is detected that there is a traffic request for the target traffic scenario, select the top target number of advertisements from the recall advertisement ranking data of the target traffic scenario to enter the recall link, where the target traffic scenario is any one of all traffic scenarios.

[0120] Optionally, in this embodiment, the above communication bus may be a PCI (Peripheral Component Interconnect, peripheral component interconnect standard) bus, or an EISA (Extended Industry Standard Architecture, extended industry standard structure) bus, etc. This communication bus can be divided into an address bus, a data bus, a control bus, etc. For the sake of representation, Figure 4 only a thick line is used to represent it in the figure, but it does not mean that there is only one bus or one type of bus. The communication interface is used for communication between the above electronic device and other devices.

[0121] The memory may include RAM, or may also include non-volatile memory, such as at least one disk memory. Optionally, the memory may also be at least one storage device located far from the aforementioned processor.

[0122] As an example, the memory 406 above may, but is not limited to, include the first computing module 302, the second computing module 304, the sorting module 306, and the recall module 308 in the above advertisement recall device. In addition, it may also include, but is not limited to, other module units in the above advertisement recall device, which will not be elaborated in this example.

[0123] The aforementioned processor may be a general-purpose processor, which may include, but is not limited to: CPU (Central Processing Unit, central processing unit), NP (Network Processor, network processor), etc.; it may also be a DSP (Digital Signal Processing, digital signal processor), ASIC (Application Specific Integrated Circuit, application-specific integrated circuit), FPGA (Field-Programmable Gate Array, field-programmable gate array), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.

[0124] Optionally, the specific examples in this embodiment may refer to the examples described in the above embodiment, and will not be elaborated here.

[0125] Those of ordinary skill in the art can understand that Figure 4 The structure shown is only for illustration. The device for implementing the above advertisement recall method may be a terminal device, which may be a smart phone (such as an Android phone, an iOS phone, etc.), a tablet computer, a personal digital assistant, and a mobile Internet device (Mobile Internet Devices, MID), a PAD and other terminal devices. Figure 4 It does not limit the structure of the above electronic device. For example, the electronic device may also include more or fewer components (such as a network interface, a display device, etc.) than those shown Figure 4 in the figure, or have a different configuration from that shown Figure 4 in the figure.

[0126] Those of ordinary skill in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by instructing the relevant hardware of the terminal device through a program, and the program can be stored in a computer-readable storage medium. The storage medium may include: a flash drive, a ROM, a RAM, a magnetic disk, or an optical disc, etc.

[0127] According to another aspect of the embodiments of the present application, there is also provided a computer-readable storage medium storing a computer program, wherein when the computer program is run by a processor, it executes the steps in the above-mentioned advertisement recall method.

[0128] Optionally, in this embodiment, those of ordinary skill in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by instructing the relevant hardware of the terminal device through a program, and the program can be stored in a computer-readable storage medium. The storage medium may include: a flash drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disc, etc.

[0129] The serial numbers of the embodiments of the present application above are only for description and do not represent the advantages or disadvantages of the embodiments.

[0130] If the integrated unit in the above embodiments is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in the above-mentioned computer-readable storage medium. Based on such an understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for causing one or more computer devices (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present application.

[0131] In the above embodiments of the present application, the descriptions of the various embodiments each have their own emphases. For the parts not detailed in a certain embodiment, reference may be made to the relevant descriptions of other embodiments.

[0132] In the several embodiments provided by the present application, it should be understood that the disclosed client can be implemented in other ways. Among them, the device embodiments described above are only illustrative. For example, the division of the units is only a logical function division, and there may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed mutual coupling or direct coupling or communication connection can be through some interfaces, and the indirect coupling or communication connection of the units or modules can be in an electrical or other form.

[0133] The unit described as a separation component may or may not be physically separated. The component shown as a unit may or may not be a physical unit, that is, it may be located in one place or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0134] In addition, each functional unit in various embodiments of the present application may be integrated into a processing unit, may exist separately as individual physical units, or two or more units may be integrated into one unit. The above-mentioned integrated units can be implemented in the form of hardware or in the form of software functional units.

[0135] The above are only the preferred embodiments of the present application. It should be noted that for those of ordinary skill in the art, without departing from the principle of the present application, several improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of the present application.

Claims

1. A method for recalling an advertisement, characterized in that: include: Calculate the historical display cost and price adjustment sensitivity coefficient of each advertisement in each traffic scenario according to the real-time bid of each advertisement in each traffic scenario; Calculating a recall score of each advertisement in each traffic scenario according to the historical display cost and the price adjustment sensitivity coefficient; Sorting all the advertisements according to the recall scores of all the advertisements in each of the traffic scenarios to obtain recall advertisement ranking data for each of the traffic scenarios; When a traffic request is detected for a target traffic scenario, the first target number of advertisements in the recall advertisement ranking data of the target traffic scenario are taken to enter the recall phase, wherein the target traffic scenario is any one of all the traffic scenarios.

2. The method according to claim 1, characterized in that Calculating the historical display cost and price adjustment sensitivity coefficient of each advertisement in each traffic scenario according to the real-time bid of each advertisement in each traffic scenario includes: Obtain the historical revenue and historical exposure times of each of the advertisements in each of the traffic scenarios; The historical display cost of each advertisement in each traffic scenario is calculated by the following formula: Among them, the CPM i,j is the historical display cost of the i-th advertisement among all the advertisements in the j-th traffic scenario among all the traffic scenarios, and the In i,j is the historical revenue of the i-th advertisement in the j-th traffic scenario, and the Im i,j is the historical exposure count of the i-th advertisement in the j-th traffic scenario.

3. The method according to claim 2, characterized in that The calculating of the historical display cost of each advertisement in each traffic scenario includes: In the case where the first advertisement has no historical revenue and historical exposure times, obtaining the historical display costs of all advertisements in a first traffic scenario, wherein the first advertisement is any one of all the advertisements, and the first traffic scenario is each traffic scenario of all the traffic scenarios; Calculate the average historical display cost in the first traffic scenario according to the historical display costs of all advertisements in the first traffic scenario; The average historical display cost is determined as the historical display cost of the first advertisement in the first traffic scenario.

4. The method according to claim 1, characterized in that: Calculating the historical display cost and price adjustment sensitivity coefficient of each advertisement in each traffic scenario according to the real-time bid of each advertisement in each traffic scenario includes: Obtaining the historical average bid and real-time bid of each of the advertisements in each of the traffic scenarios; The price adjustment sensitivity coefficient of each advertisement in each traffic scenario is calculated by the following formula: Among them, the K i,j is the price adjustment sensitivity coefficient of the i-th advertisement among all the advertisements in the j-th traffic scenario among all the traffic scenarios, and the bid_new i,j is the real-time bidding of the i-th advertisement in the j-th traffic scenario, the bid_old i,j is the historical average bid of the i-th advertisement in the j-th traffic scenario.

5. The method according to claim 1, characterized in that Calculating the recall score of each advertisement in each traffic scenario according to the historical display cost and the price adjustment sensitivity coefficient includes: The recall score of each advertisement in each traffic scenario is calculated by the following formula: S i，j =CPM i，j *K i，j i=1、2…m,j=1、2…n; Among them, the S i，j is the recall score of the i-th advertisement among all the advertisements in the j-th traffic scenario among all the traffic scenarios, and the CPM i，j is the historical display cost of the i-th advertisement in the j-th traffic scenario, and K i，j is the price adjustment sensitivity coefficient of the i-th advertisement in the j-th traffic scenario.

6. The method according to claim 1, characterized in that The step of sorting all the advertisements according to the recall scores of all the advertisements in each traffic scenario to obtain the sorting data of recalled advertisements in each traffic scenario includes: Take each of the traffic scenarios as the current traffic scenario, and perform the following operations on the current traffic scenario: Obtaining recall scores of all the advertisements in the current traffic scenario; All the advertisements are sorted according to the order of the recall scores from large to small to obtain the recalled advertisement sorting data of the current traffic scenario.

7. The method according to claim 1, characterized in that After obtaining the recall advertisement ranking data under each of the traffic scenarios, each of the recall advertisement ranking data is saved in the target database, and when a traffic request is detected for the target traffic scenario, the first target number of advertisements in the recall advertisement ranking data of the target traffic scenario are taken to enter the recall phase, including: Acquire recall advertisement ranking data of the target traffic scenario from the target database; The first target number of advertisements in the recall advertisement ranking data of the target traffic scenario are taken to enter the recall phase.

8. An advertisement recall device, characterized in that: include: A first calculation module is used to calculate the historical display cost and price adjustment sensitivity coefficient of each advertisement in each traffic scenario according to the real-time bid of each advertisement in each traffic scenario; A second calculation module, configured to calculate a recall score of each advertisement in each traffic scenario according to the historical display cost and the price adjustment sensitivity coefficient; A sorting module, used to sort all the advertisements according to the recall scores of all the advertisements in each of the traffic scenarios, and obtain the recalled advertisement sorting data of each of the traffic scenarios; The recall module is used to take the first target number of advertisements in the recall advertisement sorting data of the target traffic scenario and enter the recall link when a traffic request is detected in the target traffic scenario, wherein the target traffic scenario is any one of all the traffic scenarios.

9. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 7 is executed.

10. An electronic device comprising a memory and a processor, characterized in that: A computer program is stored in the memory, and the processor is configured to execute the method according to any one of claims 1 to 7 through the computer program.

Images