Disclosure of Invention
In order to solve the above problems, an aspect of the present invention provides a method for controlling advertisement inventory, including: counting flow data of each advertisement in the inventory; estimating the flow value distribution of each advertisement in a target time period by using the flow data; and according to the flow value distribution, deciding the exposure of each advertisement.
Further, the traffic data of each advertisement includes exposure and click rate of each advertisement.
Further, the estimating, by using the traffic data, the traffic value distribution of each advertisement in the target time period specifically includes: and estimating the flow value distribution of each advertisement in the target time period by utilizing the corresponding relation of the statistical historical flow data in time.
Further, the deciding the exposure of each advertisement according to the traffic value distribution includes: obtaining the flow value of one moment according to the flow value distribution of each advertisement on the target time period; exposing advertisement decisions higher than a preset value and not exposing advertisement decisions lower than the preset value for the flow value of each advertisement at the moment.
Further, the preset value is calculated by the following formula:
wherein epv is stock and pv is total flow.
Further, the flow value distribution of each advertisement in the target time period is estimated by the following steps:
wherein Si is the preset value, S is the flow value of the advertisement at a moment,
Count (t.gtoreq.T, d.gtoreq.D-N) is the total flow after the current hour of the history N days,
Count (S > -i, t.gtoreq.T, d.gtoreq.D-N) is the traffic of the advertisement with traffic value higher than Si after the current hour of the history N days.
Optionally, before counting the traffic data of each advertisement in the inventory, the method includes: carrying out advertisement retrieval to obtain advertisement retrieval results, user information and context information; and acquiring the traffic value of the advertisement according to the advertisement retrieval result, the user information and the context information. Specifically, a search is performed for a certain advertisement to obtain a search result, user information, and context information. The search result includes the click rate of the advertisement, the total value of generated economic benefit or the conversion rate within the range of the specified click times, and the like, for example, the higher the click rate is, the higher the propaganda strength of the advertisement is, the higher the total value of economic benefit or the conversion rate is, the higher the conversion rate of the economic value of the advertisement is, and the higher the traffic value of the advertisement is. The user information includes user group information, such as for game advertisements, that the younger user group is generally considered to have a higher continuous degree of attention than the older user group; for cosmetics, for example, the higher the conversion rate of female user population than male user population, the higher the traffic value of the corresponding advertisement. The context information may include the number of clicks and browsing time of a user on an advertisement, and the more clicks and longer browsing time indicate that the higher the interest level of the user in the advertisement, the higher the traffic value of the advertisement.
Optionally, the method for controlling advertisement inventory further includes an experiment management method, where the experiment management method includes: configuring advertisement inventory of an experiment group and a control group; counting advertisement exposure of an experimental group and a control group; and according to a strategy iteration method, when the advertisement exposure of the experimental group reaches the upper limit of inventory, screening out the advertisements with the flow value higher than the preset value for exposure.
Still another aspect of the present invention provides a control apparatus for an effect advertisement inventory, including: the system comprises a processor and a memory, wherein the memory stores a computer program, and the processor runs the computer program to realize the control method of the effect advertisement inventory.
Yet another aspect of the invention provides a computer-readable storage medium having stored thereon a computer program which, when executed by a processor, implements a method as claimed in any of the preceding claims.
According to the above, the advertisement inventory control method of the present invention includes: counting flow data of each advertisement in the inventory; estimating the flow value distribution of each advertisement in a target time period by using the flow data; and screening out advertisements for exposure according to the flow value distribution. Through deciding whether each advertisement is exposed or not in a targeted way, advertisements with high value flow can be screened out for exposure under limited inventory, and the beneficial effect of improving the rendering efficiency is achieved.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, embodiments of the present invention are described in further detail below.
The advertisement inventory control method of the invention comprises the following steps: counting flow data of each advertisement in the inventory; estimating the flow value distribution of each advertisement in a target time period by using the flow data; and according to the flow value distribution, deciding the exposure of each advertisement. Through deciding whether each advertisement is exposed or not in a targeted way, advertisements with high value flow can be screened out for exposure under limited inventory, and the beneficial effect of improving the rendering efficiency is achieved.
The traffic value of the advertisement can be understood as an index for measuring the effect of the advertisement. The effectiveness of the advertisement is determined by using scientific methods to identify the economic, social and psychological benefits of the advertisement. The economic benefit of the advertisement refers to the degree to which the advertisement campaign promotes commodity sales or labor sales and profit increases; the social benefit of the advertisement refers to the social education effect; the psychological benefit of the advertisement mainly refers to the psychological response degree of the advertisement on consumers, and brand impressions established by products can be finally promoted to purchase. In the process of detecting the effect of the advertisement by the platform, key indexes such as exposure quantity, click rate, attention quantity, click average price, total expense and the like are required to be paid attention to, and the flow value of the advertisement is obtained through the indexes so as to measure the effect of the advertisement, in other words, if the flow value is high, the good benefit of the advertisement is indicated. Therefore, in this embodiment, some advertisements in the inventory are screened out and exposed according to the flow value.
Referring to fig. 1, a flowchart of a method for controlling advertisement inventory according to an embodiment of the invention includes the following steps:
s1: counting flow data of each advertisement in the inventory; s2: estimating the flow value distribution of each advertisement in a target time period by using the flow data;
S3: and according to the flow value distribution, deciding the exposure of each advertisement.
The purpose of the step S1 is to count flow data corresponding to each advertisement in a certain historical time. For example, to estimate the traffic value of each advertisement at the current time using step S2, a previous history may be selected for a few days to make statistics. And then, the exposure of each advertisement is decided by estimating the flow value distribution of each advertisement in the target time period, so that the advertisement with high flow value can be screened out for exposure, and the rendering efficiency is improved.
Further, the traffic data of each advertisement includes exposure and click rate of each advertisement, and the traffic value is obtained by using the advertisement profits corresponding to the data, for example, the higher the profit Ecpm value of thousands of advertisement exposures is, the higher the traffic value is indicated, and the traffic value= max Ecpmi i=1..n is expressed, wherein i is a positive integer and represents each advertisement, so that the advertisement with high traffic value can be found.
Further, the estimating, by using the traffic data, the traffic value distribution of each advertisement in the target time period specifically includes: and estimating the flow value distribution of each advertisement in the target time period by utilizing the corresponding relation of the statistical historical flow data in time. In particular, there are multiple advertisements in inventory, for each advertisement, the traffic value at different times may be different; in other words, from a temporal perspective, advertisements with high traffic value may be screened out differently at different times. In summary, the traffic value distribution of each advertisement at the target time can be estimated by the statistical correspondence of the historical traffic data of each advertisement over time. For example, there are advertisement A and advertisement B in inventory, and users have historically liked to click on advertisement A and generate transactions on Monday morning, mostly ignored for advertisement B; and on Tuesday, advertisement A is rarely clicked, advertisement B is favored to be clicked and a transaction is generated, so that the flow value distribution of advertisement A and advertisement B in time of the current week is estimated by using the statistical historical flow data, and the value performance of advertisement A and advertisement B in different times can be clearly known.
Optionally, fig. 2 is a schematic flow chart of step S3 in fig. 1, including the following steps:
s31: obtaining the flow value of one moment according to the flow value distribution of each advertisement on the target time period;
s32: exposing advertisement decisions higher than a preset value and not exposing advertisement decisions lower than the preset value for the flow value of each advertisement at the moment.
For example, according to the estimated flow value distribution of the advertisement A and the advertisement B in the current week, it is known that the flow value of the advertisement A in Monday morning is high, and the flow value of the advertisement A in the week is low; advertisement B has low traffic value on monday morning and high traffic value on monday. Then, from the aspect of advertisement inventory control, advertisement A can be screened out to expose on Monday morning according to that the A traffic value of Monday morning advertisements is high, and B advertisements are not exposed; similarly, advertisement B is screened out for exposure on Tuesday and not for advertisement A. In this way, for example, the advertisement to be exposed can be flexibly adjusted at various times in accordance with the consumption habit or preference of the user for the advertisement, so as to control the advertisement inventory, and the conversion of the advertisement can be promoted, thereby improving the rendering efficiency.
Further, the preset value in the step S32 may be calculated by the following formula:
Wherein epv is stock, pv is total flow,
And defining the advertisement with the flow value higher than the preset value at a certain moment as the advertisement with the high flow value.
Further, the flow value distribution of each advertisement in the target time period is estimated by the following steps:
wherein Si is the preset value, S is the flow value of the advertisement at a moment,
Count (T is greater than or equal to T, D is greater than or equal to D-N is the total flow after the current hour of the history N days,
Count (S > Si, T > T, D > D-N is the traffic of advertisements with traffic value higher than Si after the current hour of the history N days.
For example, when N is 7, the flow value distribution started in the current hour is estimated according to the historical data, and the estimation method is as follows:
wherein Si is the preset value, S is the flow value of the advertisement at a certain moment from the current hour,
Count (t.gtoreq.T, d.gtoreq.D-7) is the total flow after the current hour for seven days in history,
Count (S > Si, t.gtoreq.T, d.gtoreq.D-7) is the flow value after seven days of history current hour is greater than that of Si.
Therefore, the flow value distribution of each advertisement in time can be intuitively displayed, and the advertisement can be conveniently screened out for exposure.
According to the above, for the exposed advertisement, one click by the user generates an advertisement request, i.e. a request to enter the advertisement playing page. Since an advertisement is usually accessed by different users at different times, relevant request data of the advertisement can be obtained through advertisement retrieval, the user group, click rate ctr, profit value Ecpm and the like of the advertisement can be analyzed, and then the advertisement is converted into flow value to be displayed. For example, the number of user groups, click rate ctr, and benefit value Ecpm are positively correlated to the level of traffic value. Therefore, the advertisement inventory control method of the present invention further includes, before the traffic data of each advertisement in the statistical inventory: carrying out advertisement retrieval to obtain advertisement retrieval results, user information and context information; and acquiring the traffic value of the advertisement according to the advertisement retrieval result, the user information and the context information. The known information can be fully utilized.
Optionally, as an optimization scheme of an embodiment of the present invention, specific parameters of advertisement inventory control need to be determined by an experiment management method, so the advertisement inventory control method further includes an experiment management method, where the experiment management method includes: configuring advertisement inventory of an experiment group and a control group; counting advertisement exposure of an experimental group and a control group; and according to a strategy iteration method, when the advertisement exposure of the experimental group reaches the upper limit of inventory, screening out the advertisements with the flow value higher than the preset value for exposure. The advertisement inventory of the small flow experiment can be managed, and the algorithm optimization work can be supported.
It should be noted that the advertisement in this embodiment may be an effect advertisement. In the prior art, in order to balance the user experience with the platform revenue, an advertisement system usually only puts advertisements (called inventory) in a certain proportion of traffic to realize the effect. In the effect advertisement system, the value of the flow is more influenced by a plurality of factors such as client budget, bid intensity, advertisement style, creative freshness, advertisement sources, user sensitivity, advertisement quality and the like, so that the embodiment makes a decision on exposure of the effect advertisement, can screen high-value flow under limited inventory, and improves the rendering efficiency.
An embodiment of the present invention provides a control device for advertisement inventory, including: the advertisement inventory control system comprises a processor and a memory, wherein the memory stores a computer program, and the processor runs the computer program to realize the advertisement inventory control method provided by the invention.
An aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, which when executed by a processor implements the method for controlling advertisement inventory provided by the present invention.
In order to make the technical contents of the present invention more clearly understood, the following examples are specifically described.
The advertisement inventory of a certain platform contains three advertisements of which the contents are cosmetics, clothes and games, the historical five-day time is selected, the flow data of the three advertisements, including the exposure and the click rate, can be counted in real time, the flow value distribution of each advertisement in the current beginning day is estimated according to the historical flow data of the five days, for example, the flow value distribution can be converted into a score representation, and the scores of the flow values of each advertisement corresponding to nine am, three pm and ten pm in the day are respectively shown in the following table 1:
table 1 shows the distribution of the scores of the traffic value of three advertisements in one embodiment
It is assumed that the preset value of 5 can be calculated by a custom setting or by the above formula (1). Then, for the flow value distribution of the advertisement of "cosmetics" in the three advertisements, the flow value score of the advertisement of "cosmetics" at nine am is 8 and is greater than the preset value 5, at this time, the advertisement of "cosmetics" is exposed, and the flow value scores of the advertisement of "cosmetics" at three pm and ten pm are 2 and 3 respectively and are lower than the preset value 5, at this time, the advertisement of "cosmetics" is not exposed. From the aspect of advertisement inventory control, the following control is performed according to the traffic value distribution of each advertisement: screening advertisements with the flow value scores greater than a preset value of 5, corresponding to nine am, and having 'cosmetics' and 'game' advertisements, and controlling advertisement inventory to expose the 'cosmetics' and the 'game' advertisements at nine am and not exposing the 'clothing' advertisements; likewise, the "clothing" advertisement is exposed at three pm, and the "cosmetic" and "game" advertisements are not exposed; likewise, the "game" advertisement is exposed at ten evening, and the "cosmetic" and "clothing" advertisements are not exposed. In this way, conversion of advertisements is facilitated, thereby improving rendering efficiency.
An embodiment of the present invention further provides a control system for advertisement inventory, as shown in part a of fig. 3, the relevant data (e.g. advertisement search result, user information and context information) after the advertisement request and the advertisement search are input to the traffic value evaluation module 010, the evaluation result is input to the exposure decision module 020 (the input of the module is the evaluation value of the traffic value evaluation module, and the output is whether to perform the advertisement exposure action), and the exposure decision module 020 outputs whether to perform the advertisement exposure action. As shown in fig. 4, after the traffic value evaluation module 010 inputs traffic data of a plurality of advertisements, the historical traffic value of each advertisement is evaluated, and the traffic evaluation score may be expressed as a traffic evaluation score according to the evaluation result, and input into the exposure decision module 020. FIG. 5 is a schematic flow chart of the operation of an exposure decision module, which performs traffic scoring distribution estimation according to historical data based on the counted advertisement exposure, and the exposure decision module 020 judges whether the advertisement is exposed according to a decision mechanism, wherein the decision mechanism is as follows:
① The lowest flow value threshold Si (preset value) is obtained from equation (1), wherein epv is the stock, and pv is the total flow.
② Decision making: if the current score S > Si, then the advertisement is exposed, otherwise no advertisement exposure is performed.
Optionally, the control system for advertisement inventory is optimized, and specific parameters of advertisement inventory control may be determined by an experiment management module, so an embodiment of the present application further provides an experiment management module, as shown in part B in fig. 3, which is an advertisement inventory for managing a small-flow experiment, and supports algorithm optimization, where the advertisement inventory for the small-flow experiment may be an advertisement inventory for performing a small-scale experiment before advertisement is put in an actual scene, and the advertisement flow is smaller at this time. Specifically, as shown in fig. 6, an experimental configuration is first performed: the flow ratio is respectively configured for the experimental group and the control group according to the configuration format and the experimental name; and then carrying out experimental flow statistics: counting the advertisement exposure of an experimental group and a control group; and then performing experimental flow control: judging whether the experimental inventory reaches the upper limit, and if so, filtering advertisements to expose by using the control method of the advertisement inventory.
According to the above, the advertisement inventory control method of the present invention includes: counting flow data of each advertisement in the inventory; estimating the flow value distribution of each advertisement in a target time period by using the flow data; and according to the flow value distribution, deciding the exposure of each advertisement. Therefore, advertisements with high value flow can be screened out for exposure under the limited inventory, and the beneficial effect of improving the rendering efficiency is achieved.
In this specification, the invention has been described with reference to specific embodiments thereof. It will be apparent that various modifications and variations can be made without departing from the spirit and scope of the invention. The specification and drawings are, accordingly, to be regarded in an illustrative rather than a restrictive sense.