CN112488760A

CN112488760A - System and method for automatically selecting non-real-time advertisements according to time dimension

Info

Publication number: CN112488760A
Application number: CN202011370466.2A
Authority: CN
Inventors: 杨威
Original assignee: Shanghai Kuliang Information Technology Co Ltd
Current assignee: Shanghai Kuliang Information Technology Co Ltd
Priority date: 2020-11-30
Filing date: 2020-11-30
Publication date: 2021-03-12

Abstract

A system for automatically selecting non-real-time advertisements according to time dimension comprises a receiving module, a price obtaining module, an adjusting module, a screening module, a bidding module and a recording module; the receiving module is used for receiving a bid request of a channel provider and acquiring the characteristic data of a target user and the request interval time of the channel provider; the price acquisition module is used for acquiring the weighted historical price of the advertisement space according to the historical price of the advertisement space; the adjusting module is used for adjusting the weighted historical price according to the request interval time of the channel provider and the change of the time, and obtaining the final price; the screening module is used for screening out advertisements to be delivered according to the directional conditions set by the advertiser and the final price; wherein the offer of the advertisement to be delivered is closest to the final price; the bidding module is used for sending the quoted price of the advertisement to be released to a channel provider for bidding; the recording module is used for recording the bidding request source, the bidding time, the price and the bidding result. The invention improves the success rate and the effect of bidding.

Description

System and method for automatically selecting non-real-time advertisements according to time dimension

Technical Field

The invention relates to the technical field of internet advertisements, in particular to a system and a method for automatically selecting non-real-time advertisements according to time dimension.

Background

In the field of internet advertisement, different advertisements compete for the same advertisement space in a bidding mode, and advertisements with high bids obtain display opportunities. Because the display opportunities and the effects of the advertisement spots are different, the advertisement spots can be divided into high-quality advertisement spots and common advertisement spots. Generally, the price of a premium ad slot is higher than that of a normal ad slot, and the price of the same ad slot floats within a certain interval. For example, an advertisement may be offered at a price that does not provide more opportunities to be presented than an advertisement that is offered at a higher price. In the case of limited budget, the advertiser can set the bid price to ensure that the advertisement is displayed for the most number of times. But for the advertiser, when a bid request from the channel provider is received, one of the advertisements is selected for bidding. Whether each bidding is successful or not directly influences the advertisement display condition of the advertiser. For example, a method for estimating bids and budgets for real-time bidding advertisements is disclosed, in which more detailed and quantitative suggestions are provided for advertisers, that is, bid suggestions and budget suggestions are provided according to the number of clicks of users that the advertisers wish to obtain, and the suggested bids and budget suggestions vary with the number of clicks that the advertisers wish to obtain. Firstly, allowing an advertiser to set the number of clicks that the advertiser wants to obtain in a release period, generally one day, then calculating a suggested bid price for obtaining the number of clicks for the advertiser according to historical data of an advertising campaign similar to the target advertising campaign, and estimating a budget required for obtaining the number of clicks according to similar historical display records, thereby providing intuitive suggestions for the advertiser to more reasonably set the budget and the bid of the advertising campaign. In the technical scheme, possible bid-winning prices of the advertisements are evaluated by historical data so as to improve the bidding success rate, but the influence of time variation of the advertising bids is not considered.

Disclosure of Invention

The technical problem to be solved by the invention is to provide a system and a method for automatically selecting non-real-time advertisements according to time dimension, which adjust the expected display price of an advertisement space by considering time factors, realize the accurate adjustment of the price of the advertisement space, improve the success rate of advertisement bidding, thus leading the advertisements to obtain more display opportunities, and perform key putting on important time intervals to improve the display effect of the advertisements.

In order to solve the above technical problem, the present invention provides a system for automatically selecting non-real-time advertisements according to a time dimension, wherein the system is characterized in that: the system comprises a receiving module, a price obtaining module, an adjusting module, a screening module, a bidding module and a recording module;

the receiving module is used for receiving a bid request of a channel provider and acquiring the characteristic data of a target user and the request interval time of the channel provider;

the price acquisition module is used for acquiring the weighted historical price of the advertisement space according to the historical price of the advertisement space;

the adjusting module is used for adjusting the weighted historical price according to the request interval time of the channel provider and the change of the time, and obtaining the final price;

the screening module is used for screening out advertisements to be delivered according to the directional conditions set by the advertiser and the final price; wherein the offer of the advertisement to be delivered is closest to the final price;

the bidding module is used for sending the quoted price of the advertisement to be released to a channel provider for bidding;

the recording module is used for recording the bidding request source, the bidding time, the price and the bidding result.

In the technical scheme, the time is adjusted according to the request interval time of the channel provider, so that the advertisement display before the next request of the channel can obtain higher bids. The request interval time of the channel provider refers to the average time interval between two requests and bids of the channel provider to the system adjacent.

As an improvement of the above scheme, the system further comprises a correlation module;

and the association module is used for analyzing the data recorded by the recording module to obtain a plurality of advertisement positions with high correlation.

In the technical scheme, the correlation of a plurality of advertisement slots is obtained by analyzing the change of key indexes such as time, price and bid results related to advertisement display, and a plurality of groups of advertisement slots with higher correlation are selected from the correlation, so that a foundation is laid for quickly responding to the price change of the advertisement slots. The bid request source refers to the ad slot that is currently bidding on. The source of the bid request, the time of the bid, the price and the result of the bid corresponding to one bid request are recorded as a record.

As an improvement of the scheme, when the bid of one advertising space changes, the adjusting module adjusts the final price of the advertising space with high relevance to update the final price.

In the technical scheme, the system is more sensitive to price response by responding to the price change relation among the advertisement positions with high correlation, so that the screened advertisements are ensured to have a greater chance to be displayed. Meanwhile, the success rate of the overall advertisement bidding of the system can be improved.

As an improvement of the above scheme, the advertisement positions with high relevance have the same user type.

In the above technical solution, the user type refers to a category into which the user is divided, such as games, e-commerce, education, beauty, food, and the like. No high correlation is formed between ad spots that do not belong to the same category. The technical scheme limits the range between the advertisement positions with high correlation, so that the correlation judgment of the advertisement positions has more practical guiding significance, and the price change between the advertisement positions with high correlation can indicate the price change conditions of other advertisement positions.

As an improvement of the above, the high correlation means that the absolute value of the correlation coefficient is 0.8 or more.

In the above technical solution, the correlation coefficient adopts a pearson correlation coefficient manner. And calculating the prices of different advertisement positions at the same time to obtain a plurality of advertisement positions with high relevance. The technical scheme has the advantages that the judgment standard of high correlation is customized, the standard capable of high-efficiency judgment is given, and the standardization of correlation judgment is realized.

As an improvement of the above, the time refers to different time periods in a day and different time ranges in a year, especially the time ranges of various types of holidays.

In the technical scheme, the change characteristics of the daily lives and work and rest of people within one day and the behavior characteristics of different time periods within one year are considered, so that the purpose of directionally popularizing the advertisement according to the requirements of the user can be accurately realized, and the advertisement display effect can be improved.

As an improvement of the above, the weighted historical prices are weighted averages of prices at different periods in a past fixed period, and the prices are weighted more heavily closer to the current time.

In the above technical solution, the past fixed period is divided into several small periods, and these small periods constitute the fixed period. Calculating the average value of the winning bid price for each small period, and then calculating the weighted average value of the average values, wherein the closer the current time is, the higher the weight of the price is. The minimum unit of the fixed period is day, so the historical price of the advertisement in one day remains unchanged.

As an improvement of the above scheme, the time is divided into a focus time and a general time according to user performance at different times, and the final bid at the focus time is higher than the final bid at the general time.

In the technical scheme, different time periods are divided according to user performance, and key time and general time are distinguished, so that price change can be more precise and accurate. Because the emphasis time has better display effect, the bid price is properly increased at the emphasis time so as to obtain a larger chance for obtaining the display opportunity. The emphasis time and the general time herein refer to general expressions of classification, and do not refer to the classification of time into two categories alone. Each category can be further subdivided into a plurality of different small categories to meet the requirement of higher fine-grained delivery.

Correspondingly, the invention also provides a method for automatically selecting the non-real-time advertisement according to the time dimension, which comprises the following steps.

A. And receiving a bid request of a channel provider by using the receiving module, and acquiring the characteristic data of the target user and the request interval time of the channel provider.

In this step, the feature data of the target user is acquired, and information is provided for acquiring advertisements meeting the requirements of advertisers in the next step.

B. And obtaining the weighted historical price of the advertisement space according to the historical price of the advertisement space by using the price obtaining module.

In this step, the advertisement slot is used as an analysis object, and the price level of the advertisement slot can be evaluated by analyzing the historical data, thereby providing a basis for the subsequent selection of the advertisement matched with the advertisement slot.

C. And adjusting the weighted historical price by using the adjusting module according to the request interval time of the channel provider and the change of the time, and obtaining the final price.

In this step, price up-regulation is performed on the important time according to the fluctuation of the price of the advertisement space around the price level of the advertisement space, so that the chance of successful display is increased. Therefore, the probability of the advertisement display can be increased, and the advertisement of the advertiser can obtain better display effect in the overall view.

D. Screening out advertisements to be delivered by using the screening module according to the directional conditions set by the advertiser and the final price; wherein the offer of the advertisement to be delivered is closest to the final price.

In this step, the screened advertisements to be delivered need to satisfy the requirements of both the targeting condition and the price. When the advertisement is mainly set with the advertisement putting condition, the targeting condition is set, namely, the advertisement is expected to be displayed to which users. And B, comparing the characteristic data of the user acquired in the step A with the orientation condition, and screening out the advertisements meeting the orientation requirement. The price of the advertisement is also fixed, and the advertisement with the closest price can be sorted. And the advertisements meeting the two conditions are the advertisements to be delivered. The step does not require the sequence of comparison orientation conditions and price screening. The number of advertisements in the advertisement pool can be determined according to the number of the advertisements in the advertisement pool.

E. And sending the quoted price of the advertisement to be released to a channel provider for bidding by using the bidding module.

In this step, the adjusted bid obtained in the previous step is taken as the final bid, and the bidding process of the present scheme is realized.

F. The recording module is used for recording the source of the bid request, the time of the bid, the price and the bid result.

In the step, the condition and the result of each bidding are recorded, and an informative and reliable data source is provided for data analysis. The more data is recorded, the more reliable the results of the data analysis.

As an improvement of the above solution, the method for automatically selecting non-real-time advertisements according to the time dimension further comprises the following steps.

G. And analyzing the data recorded by the recording module by using the association module to obtain a plurality of advertisement positions with high correlation.

In this step, through the analysis of the historical data, a plurality of advertisement positions with high relevance are obtained, so that the price variation of the advertisement positions is more predictable, the bid price is more refined, and the method is a supplement of a scheme for adjusting the bid price according to the time variation.

The invention has the following beneficial effects.

The invention adjusts the expected display price of the advertisement space by considering the influence of time on the user behavior, realizes the fine prejudgment of the price of the advertisement space, and improves the success rate of advertisement bidding, thereby enabling the advertisement to obtain higher display ratio. The invention takes the highest bid price in a period of time as the final bid price, thereby ensuring the display success rate of the advertisement. Meanwhile, key putting is carried out in the important time intervals, the display effect of the advertisement is improved, and better user coverage and propaganda effect are realized. According to the invention, through selection of different times, the function of distinguishing the users is achieved, and the dimension of positioning the users is increased, so that the users can be positioned more accurately. The invention carries out targeted adjustment according to the characteristics of the user behavior, reduces the dependence on historical data and improves the adaptability of the system.

Drawings

Fig. 1 is a schematic structural diagram of a first embodiment of the system for automatically selecting non-real-time advertisements according to a time dimension.

FIG. 2 is a schematic diagram of time classification according to the present invention.

Fig. 3 is a schematic structural diagram of a second embodiment of the system for automatically selecting non-real-time advertisements according to the time dimension.

FIG. 4 is a flow diagram of a first embodiment of a method of the present invention for automatically selecting non-real-time advertisements based on a time dimension.

FIG. 5 is a flow diagram of a second embodiment of a method of the present invention for automatically selecting non-real-time advertisements based on a time dimension.

Detailed Description

In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention will be described in further detail with reference to the accompanying drawings.

In a first embodiment of the present invention, as shown in fig. 1, a system for automatically selecting non-real-time advertisements according to a time dimension is provided, which includes a receiving module 100, a price obtaining module 200, an adjusting module 300, a filtering module 400, a bidding module 500, and a recording module 600.

The receiving module 100 is configured to receive a bid request of a channel provider, and obtain feature data of a target user and a request interval time of the channel provider.

Specifically, the system simultaneously connects a plurality of channels, and feeds back bids to the channels by receiving bid requests of the channels. And the channel trader or a plurality of channel traders bid on the display of the same advertisement slot to display the advertisement with the highest bid, thereby completing the process of advertisement display. The receiving module is used for receiving a bid request of a channel provider and simultaneously comprises characteristic data of a target user, such as the size of an advertisement space, an IP address, browser information, an operating system, gender, country or region and other various information. If the channel does not have the feature data attached when the channel requests the bid, the receiving module 100 sends a request to the channel to obtain the feature data. The time for requesting the bid by different channels is different, some are 15 minutes, some are 30 minutes, and other various time such as 1 hour, but the time interval for requesting the bid by the same channel is generally fixed. The receiving module 100 may acquire bid request interval times of respective channels. If the two requests of the same distributor suddenly change greatly, the receiving module 100 regards this as an exception and still delivers the previous interval to the adjusting module. And if the request time of two continuous times is greatly changed and the same, judging that the request time is modified by the channel provider, and taking the new interval time as the request and bid interval time of the channel provider. The aforementioned large variation means that the time interval is more than 30% from the original time interval.

The price obtaining module 200 is configured to obtain a weighted historical price of the ad slot according to the historical price of the ad slot.

Specifically, the historical price of the ad slot refers to the winning bid price, i.e., the bid was shown. The weighted historical price is a weighted average of prices at different time periods over a past fixed period, and the closer to the current time, the more weight the price is. For example, the price of an advertisement delivery platform acquired by the price acquisition module 200 is 6.5 yuan, 6.1 yuan before one day, 5.5 yuan before 30 days, and 7.2 yuan before one year, and weighted values of 0.7, 0.18, 0.09, and 0.03 are respectively assigned according to time intervals, so that a weighted price of 6.359 yuan is calculated. When the weight value of the price in each time period is calculated, the current calculation is determined according to the current time of the acquired price and the current bidding request period. The weight values are rounded to two decimal places. The closer the bid time is to the current, the greater its weight.

And the adjusting module 300 is configured to adjust the weighted historical price according to the request interval time of the channel provider and the change of the time, and obtain a final price.

Specifically, people have certain regularity in life and work, so that the response of users to advertisements also presents certain regularity. For example, because people are busy in work and study in the daytime and in a tense state, even if there is an interested advertisement, the people may not click, so that various data values are low, and when the people finish work or study at night, the people have a lot of time and are in a relaxed state, so that the attraction of the advertisement to people is increased, and data is increased. Similarly, on weekends, people's status is different from weekdays, which also results in an increase in the attractiveness of the advertisement. More typically, when the national day comes on a long holiday, people can make a travel plan in advance and pay attention to products such as travel navigation and game accommodation, so that the advertisement data of the two industries are obviously increased before the long holiday. The adjustment module 300 adjusts different advertising bids as a function of time. Time is divided into a time-of-focus and a general time in consideration of user performance at different times, and a final bid at the time-of-focus is higher than a final bid at the general time. The emphasis time is further subdivided into emphasis type I time and emphasis type II time. The general time is further subdivided into general class I time and general class II time. Depending on the time of focus classification, the default bid is floated upwards by a different scale. The default bid is floated down a different scale according to a general temporal classification. For example, for travel-type advertisements, as shown in FIG. 2, all times are marked as important dates some time before five or national day, and peak times within one day are also marked as important. During the peak of the five or national day celebrations, the days outside the peak of these days are marked as important class I, while the days outside the peak of these days are marked as important class II. The peak periods in the general days are labeled as general class I, and the times other than the peak periods in the general days are labeled as general class II. Of course, non-holiday weekends may also be considered, with additional categories marking and bid adjustment. Because the general ideas are consistent, the details are not repeated herein. If a channel has an interval request time of 15 minutes, the critical time is 19: 00-21: 00, when the system is in 18: 50, when receiving the bid request of the channel, the effective time for obtaining the bid is 18: 50-19: 05, at 18: 50-19: 00, is general time, and 19: 00-19: since 05 is a significant time, the bid for that time is adjusted to the bid for the significant time.

A screening module 400, configured to screen out an advertisement to be delivered according to a targeting condition set by an advertiser and the final price; wherein the offer of the advertisement to be delivered is closest to the final price.

Specifically, when an advertiser configures an advertisement, targeted client information of the advertisement is configured, and various information such as a region or a country of a user, a browser, an operating system, a gender, and the like are set. By receiving the specific data obtained by the module 100, the advertisement can be reversely screened out to meet the requirements of which advertisements, so that the advertisement to be delivered is limited within a certain range. The system is specific to various types of information of the advertisement, including information of bidding, budget and the like. The screening module 400 selects the advertisement with the price closest to the final price from the range as the advertisement to be delivered.

And the bidding module 500 is used for sending the quote of the advertisement to be released to a channel provider for bidding.

In particular, the bidding module 500 completes bidding for an advertisement. In bidding, the bidding module 500 need only send the price of the advertisement to the channel, and need not send the advertisement content to the channel altogether.

A recording module 600 for recording the source of the bid request, the time of the bid, the price and the bid result.

Specifically, after one bidding is completed, the recording module 600 records the time, price and bidding result of the current bidding. The bidding results are divided into two categories, success and failure. The recording module stores the recorded data in the database in a structured storage manner, which facilitates quick retrieval and analysis by the association module 500.

In a second embodiment of the present invention, as shown in fig. 3, a system for automatically selecting non-real-time advertisements according to a time dimension is provided, which further comprises an association module 700, compared to the first embodiment;

the association module 700 is configured to analyze the data recorded by the recording module to obtain a plurality of advertisement slots with high relevance.

Specifically, different ad slots may have a certain correlation due to the same user group and the association of websites. And the price of part of the ad slots also shows certain relevance due to high relevance, so that the method has great significance for bid prediction. The correlation coefficient is a statistical index used for reflecting the degree of closeness of the correlation between variables, and the invention adopts the Pearson correlation coefficient to carry out correlation system calculation on the bid winning prices of two advertisement slots at the same time. The correlation coefficient is calculated according to a product difference method, and the degree of correlation between two variables is reflected by multiplying the two dispersion differences on the basis of the dispersion difference of the two variables and the respective average value. The advertisement positions with high correlation have the same user type, so when the correlation of the advertisement positions is judged, the advertisement positions with the same user type only need to be judged. When the absolute value of the correlation coefficient is 0.8 or more, it is determined that the correlation is high. For a plurality of advertisement slots with high relevance, if the price of one advertisement slot floats upwards, namely the price is not in the middle after the offer, and then the advertisement slot with high relevance requests the offer, the adjusting module 300 adjusts the offer upwards, so that the screening module can change the screened advertisements.

Accordingly, a third embodiment of the present invention, as shown in FIG. 4, provides a method for automatically selecting non-real-time advertisements based on a time dimension, comprising the following steps.

S1, receiving a bid request of a channel provider by using the receiving module, and acquiring characteristic data of a target user and request interval time of the channel provider.

In this step, receiving a bid request from a channel provider is the first step of the method. The method begins operation in response to the bid request and receives target user characteristic data that may be used to screen advertisements, providing a basis for screening advertisements in step S4.

And S2, obtaining the weighted historical price of the advertisement space according to the historical price of the advertisement space by using the price obtaining module.

In this step, the weighted historical price of the advertisement space is obtained by calculating the historical price of the advertisement space and is used as the base number for estimating the transaction price of the advertisement space. This step is obtained from historical data recorded in the system. This step is also performed after step S1, making the acquisition of data more targeted. Compared with a simple arithmetic mean value, the method adopting the weighted historical price is more sensitive to the current overall price trend, and meanwhile, systematic changes caused by economic environmental changes can be filtered out.

And S3, adjusting the weighted historical price by using the adjusting module according to the request interval time of the channel provider and the change of the time, and obtaining the final price.

In this step, the weighted historical prices in step S2 are adjusted according to the current time characteristics, so that price floating is realized. Such adjustments can ensure that bids are placed at higher prices during key-points, thereby making display opportunities more accessible, while bids are placed at lower prices during general times, saving advertiser budget. The adjustment method for the historical prices has been described in the previous embodiments, and is not described herein.

S4, screening out advertisements to be delivered according to the targeting conditions set by the advertisers and the final price by using the screening module; wherein the offer of the advertisement to be delivered is closest to the final price.

In this step, under the condition that the price of the advertisement is ensured to be not moved, the advertisement with the price closest to the final price is screened out, and the price set by the advertiser is used for bidding, so that the setting of the advertiser is realized, and the display can be realized to the greatest extent.

And S5, sending the quote of the advertisement to be placed to a channel for bidding by using the bidding module.

In this step, the price of the advertisement to be delivered obtained in step S4 is transmitted to the distributor as a bid price, thereby achieving the final bid.

S6, recording the source of the bid request, the time of the bid, the price and the result of the bid by using the recording module.

In this step, the bid information and result in step S5 are recorded and stored in the database in a structured storage manner, so that the speed of step S2 can be increased, thereby improving the operation efficiency of the present embodiment.

A fourth embodiment of the present invention, as shown in fig. 5, provides a method for automatically selecting non-real-time advertisements according to a time dimension, which further includes the following steps compared to the third embodiment.

And S7, analyzing the data recorded by the recording module by using the association module to obtain a plurality of advertisement spots with high relevance.

In this step, through the analysis of the historical data recorded in the database, a plurality of advertisement slots with high correlation are obtained, so that the adjustment of the bid according to the time in the previous embodiment is supplemented, the sensitivity of the bid is improved, and the change of the price is responded to more quickly. The calculation method of the correlation has been described in the previous embodiments, and is not described herein again.

While the invention has been described in connection with what is presently considered to be the most practical and preferred embodiment, it is to be understood that the invention is not to be limited to the disclosed embodiment, but on the contrary, is intended to cover various modifications and equivalent arrangements included within the spirit and scope of the appended claims.

Claims

1. A system for automatically selecting non-real time advertisements based on a time dimension, comprising: the system comprises a receiving module, a price obtaining module, an adjusting module, a screening module, a bidding module and a recording module;

2. The system for automatically selecting non-real-time advertisements based on a time dimension of claim 1 further comprising an association module;

3. The system of claim 2, wherein the adjustment module adjusts the final price of a slot of high relevance to update the final price as bids for the slot change.

4. A system for automatically selecting non-real time advertisements based on a time dimension as recited in claim 2, wherein the highly relevant ad slots have the same user type.

5. The system of claim 2, wherein the high correlation is a correlation coefficient having an absolute value of 0.8 or more.

6. A system for automatic selection of non-real time advertisements based on a time dimension as claimed in claim 1 wherein the time refers to both different time periods of the day and different time ranges of the year, especially of various types of holidays.

7. The system of claim 1, wherein the weighted historical prices are weighted averages of prices at different time periods over a past fixed period, and wherein prices are weighted more heavily closer to the current time.

8. A system for automatically selecting non-real time advertisements based on a time dimension as recited in claim 1, wherein said time is divided into a focus time and a general time based on user performance at different times, and wherein said final bid at said focus time is higher than said final bid at said general time.

9. A method for automatically selecting non-real-time advertisements based on a time dimension, comprising:

A. receiving a bid request of a channel provider by using the receiving module of any one of claims 1-8, and acquiring feature data of a target user and a request interval time of the channel provider;

B. obtaining a weighted historical price for an ad placement from a historical price for the ad placement using a price acquisition module of any of claims 1-8;

C. adjusting the weighted historical prices according to the request interval time and the change of time of the channel using the adjusting module of any one of claims 1-8 and obtaining the final price;

D. screening out advertisements to be delivered according to the targeting conditions set by the advertisers and the final price by using the screening module of any one of claims 1-8; wherein the offer of the advertisement to be delivered is closest to the final price;

E. sending the offer for the advertisement to be placed to a channel for bidding using the bidding module of any one of claims 1-8;

F. recording a bid request source, a time of bid, a price, and a bid result using the recording module of any of claims 1-8.

10. The method of claim 9, wherein automatically selecting a non-real-time advertisement based on a time dimension further comprises:

G. the data recorded by the recording module is analyzed by using the correlation module as claimed in claim 2 to obtain a plurality of ad spots with high correlation.