CN106296247B - Network information resource online ordering method and device - Google Patents
Network information resource online ordering method and device Download PDFInfo
- Publication number
- CN106296247B CN106296247B CN201510276519.7A CN201510276519A CN106296247B CN 106296247 B CN106296247 B CN 106296247B CN 201510276519 A CN201510276519 A CN 201510276519A CN 106296247 B CN106296247 B CN 106296247B
- Authority
- CN
- China
- Prior art keywords
- network information
- exposure
- estimated
- under
- orientation type
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention relates to a network information resource online sequencing method and device. The method comprises the following steps: acquiring estimated exposure and actual exposure of network information; obtaining an estimated error proportion under the orientation type according to the estimated exposure and the actual exposure; correcting the pre-acquired estimated crowd quantity according to the estimated error proportion under the orientation type to obtain the corrected estimated crowd quantity under the orientation type; calculating the exposure probability of the network information under the directional type according to the corrected estimated population quantity; acquiring expected income and expected income distribution probability density of network information exposure under the orientation type; acquiring a bias parameter of the network information under the orientation type according to the exposure probability of the network information under the orientation type and the corresponding expected revenue distribution probability density; and carrying out online sequencing on the network information according to the expected income and the bias parameters. The accuracy of resource allocation is improved.
Description
Technical Field
The invention relates to the field of information processing, in particular to a network information resource online sequencing method and device.
Background
With the development of internet technology, the internet provides various network information services. The network information service may include an advertisement service, an article information presentation service, and the like. The network information service, which is exemplified by an internet advertisement service, provides various advertisement forms, and can be divided into effect advertisements and presentation advertisements according to a billing manner. A large portion of internet-presented advertisements are sold by contracts, i.e., an agreement is made between the advertising platform and the advertiser to sell a particular time period, a particular user, and a particular amount of presentation.
The traditional advertisement resource allocation algorithm adopts an HWM (high Water Mark) algorithm which is divided into an off-line stage and an on-line stage, service priority is allocated to contract advertisements according to the difficulty of meeting the advertisement resources, then allocation ratio is calculated for the advertisements according to the service priority in sequence, the HWM algorithm requires that the exposed crowd is estimated accurately, the precondition is difficult to meet, for example, special dates (such as weekends, holidays and the like) and the situation of sudden change of flow of a media side are met, the number of the crowd in the day is difficult to be estimated accurately in advance, and accordingly, the allocation is inaccurate.
Disclosure of Invention
Therefore, it is necessary to provide an online network information resource sequencing method for solving the problem that the traditional network information resource allocation algorithm is difficult to allocate accurately, so as to improve the accuracy of resource allocation.
In addition, it is necessary to provide an online network information resource sorting apparatus, which can improve the accuracy of resource allocation.
A network information resource online sequencing method comprises the following steps:
acquiring estimated exposure and actual exposure of network information;
obtaining an estimated error proportion under the orientation type according to the estimated exposure and the actual exposure;
correcting the pre-acquired estimated crowd quantity according to the estimated error proportion under the orientation type to obtain the corrected estimated crowd quantity under the orientation type;
calculating the exposure probability of the network information under the directional type according to the corrected estimated population quantity;
acquiring expected income and expected income distribution probability density of network information exposure under the orientation type;
acquiring a bias parameter of the network information under the orientation type according to the exposure probability of the network information under the orientation type and the corresponding expected revenue distribution probability density;
and carrying out online sequencing on the network information according to the expected income and the bias parameters.
An online ranking device for network information resources, comprising:
the exposure acquisition module is used for acquiring the estimated exposure and the actual exposure of the network information;
the estimated error proportion acquisition module is used for acquiring an estimated error proportion under the orientation type according to the estimated exposure and the actual exposure;
the correction module is used for correcting the pre-acquired estimated crowd quantity according to the estimated error proportion under the orientation type to obtain the corrected estimated crowd quantity under the orientation type;
the exposure probability calculation module is used for calculating the exposure probability of the network information under the orientation type according to the corrected estimated population quantity;
an expected income acquisition module, configured to acquire an expected income and an expected income distribution probability density of the network information exposure in the orientation type;
the bias parameter acquisition module is used for acquiring bias parameters of each piece of network information under the orientation type according to the exposure probability of the network information under the orientation type and the corresponding expected revenue distribution probability density;
and the sequencing module is used for carrying out online sequencing on the network information according to the expected income and the bias parameters.
The online ordering method and the device for the network information resources obtain the estimated error proportion through the estimated exposure and the actual exposure of the network information, correct the estimated quantity of people according to the estimated error proportion, obtain the exposure probability of the network information according to the estimated quantity of people, obtain the offset parameter of the network information according to the expected income distribution probability density and the exposure probability, perform online ordering on the network information according to the expected income and the offset parameter, improve the accuracy of the estimated quantity of people due to the correction on the estimated quantity of people, further calculate the exposure probability and the offset parameter of the obtained network information more accurately, order the network information according to the expected income and the offset parameter, maximize the exposure effect of the network information, improve the accuracy of resource allocation, and consider the estimated click rate of the network information for the exposure of each piece of network information, the network information selects the user with the highest click rate from the potential users, so that the overall click rate of the network information is improved.
Drawings
FIG. 1 is a schematic diagram of an internal structure of a server operating a method and apparatus for online ranking of network information resources in one embodiment;
FIG. 2 is a block diagram of network information resource allocation;
FIG. 3 is a flowchart illustrating a method for online ranking of network information resources according to an embodiment;
FIG. 4 is a schematic diagram of two adjacent time corrections;
fig. 5 is a diagram illustrating a network information distribution ratio;
FIG. 6 is a schematic diagram of an ECPM distribution;
fig. 7 is a schematic structural diagram of an online ranking apparatus for network information resources in an embodiment.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
Fig. 1 is a schematic diagram of an internal structure of a server operating a network information resource online ranking method and apparatus in an embodiment. As shown in fig. 1, the server includes a processor, a storage medium, a memory, and a network interface connected by a system bus. The storage medium of the server stores an operating system, a database and a network information resource online sequencing device, wherein the database stores network information resources, and the network information resource online sequencing device is used for realizing a network information resource online sequencing method suitable for the server. The processor of the server is used for providing calculation and control capacity and supporting the operation of the whole server. The memory of the server provides an environment for the operation of the network information resource online sequencing device in the storage medium. The network interface of the server is used for communicating with an external terminal through network connection, such as receiving a display request sent by the terminal and returning network information resources to the terminal. The server may be implemented as a stand-alone server or as a server cluster consisting of a plurality of servers. Those skilled in the art will appreciate that the architecture shown in fig. 1 is a block diagram of only a portion of the architecture associated with the subject application, and does not constitute a limitation on the servers to which the subject application applies, as a particular server may include more or less components than those shown, or may combine certain components, or have a different arrangement of components.
Fig. 2 is a schematic diagram of a framework for network information resource allocation. As shown in fig. 2, the framework for the allocation of network information resources includes two phases, offline and online. The off-line stage comprises a booking module (an order module) and an allocation module (a distribution module), wherein the order module is directly butted according to a network information delivery end and is used for performing inventory estimation and cost estimation and finally forming a contract order. The contract order includes necessary information required for delivering network information resources, such as the type of orientation (for example, age 20 to 25 years, male) of the network information to be purchased by the user, the number of exposures to be purchased (for example, 100 ten thousand times a day), and the location of purchase (browser side). The distribution module performs flow distribution according to the exposure purchased by the network information owner and the estimated crowd quantity, and calculates a distribution scheme off line. The distribution engine of the distribution module inputs the contract order, the estimated exposure and the real-time exposure of the network information into the distribution scheme generator, periodically calculates, outputs the bias parameters and the statistical CPC (Cost Per Click) of the network information, serves as the distribution scheme of the network information, and pushes the distribution scheme to the network information server for online network information resource playing. And the Server (service module) of the network information server loads the offset parameter and the statistical CPC of the network information, and sorts the network information by taking the product and the offset parameter of the statistical CPC and the estimated click rate of the network information as the sorting factor of the network information, so that the network information is displayed according to the sorting result.
The network information is an offline contract, and a network information owner purchases a network information form of a specific time period, a specific crowd, a specific exposure frequency, a display charge per thousand times and an online release of the network information platform. The network information may be advertisement information, article information, news information, event information, etc. The advertisement information may be advertisement information displayed on a network platform, etc. The article information may be information of a commodity displayed on a shopping platform, or the like. The news information can be news information searched by a search engine or news information displayed on a news information publishing platform, and the like. The activity information can be activity information promoted on the network platform, such as expropriation promotion information, evaluation promotion information, micro-movie promotion information and the like.
The bias parameter is used to adjust the exposure of the network information.
Fig. 3 is a flowchart illustrating an online ranking method of network information resources according to an embodiment. The network information resource online ordering method in fig. 2 operates under the framework in fig. 2. As shown in fig. 3, an online ranking method for network information resources includes the following steps:
and step 302, acquiring the estimated exposure and the actual exposure of the network information.
In one embodiment, the step of obtaining the estimated exposure of the network information comprises: counting historical display times within preset time under the orientation type; and estimating the estimated exposure of the network information according to the historical display times.
Specifically, the targeting type refers to a category under a targeting crowd preset in the network information, such as 1 year old, 2 years old or 1 to 6 years old for a male. The predetermined time may be set as desired, such as 5 days, 7 days, or one month. The historical display times in the preset time can be used as the estimated exposure of the network information, or the historical display times in the preset time can be weighted and averaged to be used as the estimated exposure of the network information, and the like.
And step 304, obtaining the estimated error proportion under the orientation type according to the estimated exposure and the actual exposure.
Specifically, the estimated error ratio may be calculated once at predetermined time intervals, which may be 5 minutes, 10 minutes, and the like. And correcting the deviation of the estimated exposure under each orientation type according to the real-time exposure and the estimated exposure of the network information server and the orientation type of the network information each time. The specific correction formula is as follows (1):
in the formula (1), ujIndicating the network information j actual exposure, djIndicating the estimated exposure, σ, of the network information j during this timeitAnd (4) representing the exposure deviation of the orientation type i at the time t, namely the estimated error proportion under the orientation type.
And step 306, correcting the pre-acquired estimated crowd quantity according to the estimated error proportion under the orientation type to obtain the corrected estimated crowd quantity under the orientation type.
In one embodiment, step 306 includes: expecting the estimated error proportion of the network information under the orientation type through a time attenuation function to obtain the total estimated error proportion under the orientation type; and correcting the pre-acquired estimated crowd quantity according to the total estimated error proportion under the orientation type to obtain the corrected estimated crowd quantity under the orientation type.
In particular, σi＝Ef(t)[σit] (2)
f(t)＝λe-λt (3)
In the formulas (2) and (3), f (t) is a time decay function, which is represented by exponential distribution, wherein the value of λ can be determined according to actual conditions, which represents the intensity of time decay, and the smaller λ represents the greater dependence on historical data, in this embodiment, λ can be 0.5; t represents the time from the present and can range from 0 to 48 in units of half an hour. SigmaiAnd (3) representing the total estimated error proportion under the orientation type i, wherein the estimated error proportion in each time period is expected by using a time attenuation function.
Fig. 4 is a schematic diagram of correction at two adjacent time instants. As shown in FIG. 4, at the time of T-1, the contract order and the estimated exposure are input into the distribution engine to obtain a distribution scheme, the online network information server distributes according to the distribution scheme to obtain the actual exposure, the actual exposure is input into the distribution engine at the time of T, the exposure deviation at the time of T is corrected, and so on.
According to the calculated total estimated error proportion sigmaiFor pre-acquisitionCorrecting the estimated population quantity, wherein the correction formula is as follows (4):
in formula (4), SupplyiRepresenting the pre-acquired estimate of the population,
indicating the estimated population after correction.
And 308, calculating the exposure probability of the network information under the orientation type according to the estimated quantity of the corrected crowd.
In one embodiment, step 308 comprises: acquiring the estimated quantity of the directional crowd meeting the network information; determining the priority of the network information according to the satisfied estimated quantity of the directional crowd; and calculating the exposure probability of the network information under the orientation type from high to low according to the priority.
For example, the number of people satisfied by the first network information is 100W (ten thousands of people) +80W, and 180W in total; the number of people satisfied by the second network information is 80W +120W, and the total number of people is 200W. The priority of the first network information is higher than the priority of the second network information.
After the priority of each piece of network information is determined, the distribution proportion, namely the exposure probability, of each piece of network information is calculated in sequence from high to low according to the priority. If the distribution ratio of the first network information is calculated to be 150/(100+80) ═ 0.833; when calculating the distribution ratio of the second network information, the number of the people distributed to the first network information is firstly removed, the left 100 (1-0.833) of the first directed people is 16.7W, the left 80 (1-0.833) of the second directed people is 13.36W, and the distribution ratio of the second network information is 0.167+ (80-13.36-0.167)/120/0.5553.
Fig. 5 is a diagram illustrating a network information distribution ratio. As shown in fig. 5, the number of predicted people of the first network information is 150W, the number of predicted people of the second network information is 80W, the number of first targeted people is 100W, the number of second targeted people is 80W, the number of third targeted people is 120W, the number of people satisfying the first network information is 180W in total, the number of people satisfying the second network information is 200W in total, the priority of the first network information is higher than that of the second network information, the allocation ratio of the first network information is calculated to be 0.833, and the allocation ratio of the second network information is 0.5553.
The specific algorithm for calculating the exposure probability of the network information is as follows:
1) initializing remaining supply r for all orientation types ii＝si；
2) For the network information j, sorting according to the priority, and traversing a) and b);
a) solving the following equation (5) to obtain alphaj；
If the solution is not available, then set alphaj＝1。
b) For all i e (j), update ri＝ri-min{ri,siαj}。
Due to riIndicating the remaining population exposure dose, which decreases each time a network message is assigned an exposure dose, so r is updated each iterationi。
In addition, the estimated exposure of the network information can be adjusted by directly using the actual exposure fed back on line, and when the actual exposure is larger than the estimated exposure, the exposure probability of the corresponding network information is reduced, otherwise, the exposure probability is increased.
And step 310, acquiring the expected income and the expected income distribution probability density of the network information exposure under the orientation type.
In one embodiment, step 310 includes: acquiring the statistical single click cost of the network information and the estimated click rate of the network information of each exposure; calculating the product of the single click cost and the estimated click rate to obtain the expected yield of each exposure of the network information; and carrying out statistics according to the expected income of each exposure of the network information to obtain the expected income distribution probability density of the network information.
Specifically, when each piece of network information just starts to be on-line, the exposure probability of the network information is directly used for on-line selection, and the sequencing exposure is carried out according to the exposure probability of the network information.
After accumulating a predetermined number of exposures and clicks, a statistical CPC (cost per click) of the network information is calculated, denoted by stat _ CPC, where:
stat_cpc＝cost/click (6)
in the formula (6), cost represents the cost in the preset time of the network information, and click represents the number of clicks in the preset time. The predetermined number of exposures may be selected to be 10000 times and the number of clicks may be 100 times.
And calculating the ECPM (expected Cost Per future expressions, which is the expected income after 1000 network information exposures) of each network information exposure, wherein the calculation formula is shown as the formula (7).
ECPM＝stat_cpc*pCtr (7)
Wherein pCtr represents the estimated click rate of the network information of the current exposure.
The expected income of each exposure of each piece of network information is counted to obtain the expected income distribution probability density f of each piece of network informationcontract(Ecpm)。
In one embodiment, step 312 includes: calculating the expected income distribution probability density of the unconventional network information under the orientation type; and calculating the bias parameters of the network information under the orientation type through an accumulative probability function according to the exposure probability of the network information under the orientation type, the expected income distribution probability density of the unconventional network information and the expected income distribution probability density of each piece of network information.
In particular, non-approximations under statistical orientation typesDetermining expected revenue distribution probability density f for network informationnon-contract(Ecpm) obtained by log statistics, such as the number of exposures of Ecpm at [ 0, 0.1, Ecpm at [ 0.1, 0.2, etc. The non-agreed network information refers to network information other than the contract.
And (3) obtaining a bias parameter bias of the network information through an accumulative probability function F, and calculating a formula as the formula (8).
F(Ecpmcontract+Ecpmnon-contract+bias)＝α (8)
In the formula (8), α represents a network information exposure probability.
In one embodiment, the step of counting the expected revenue distribution probability density of the non-agreed network information under the directive type comprises: and counting the expected profit distribution probability density of the unconventional network information under the directional type from the log.
And step 314, performing online sequencing on the network information according to the expected income and the bias parameters.
And taking the expected income and the bias parameters as sorting factors to sort the network information.
FIG. 6 is a schematic diagram of an ECPM distribution. As shown in fig. 6, the ECPM distribution of the network information is translated by bias so that the exposure probability in the natural competition of the network information in the ranking is α, thereby achieving the purpose of keeping the amount. The ECPM distribution 61 is the overall ECPM distribution, the ECPM distribution 62 is the network information, and the ECPM distribution 63 is the network information adjusted ECPM distribution.
The online ordering method of the network information resources obtains the estimated error proportion through the estimated exposure and the actual exposure of the network information, corrects the estimated quantity of the crowd according to the estimated error proportion, obtains the exposure probability of the network information according to the estimated quantity of the crowd, obtains the offset parameter of the network information according to the expected income distribution probability density and the exposure probability, carries out online ordering on the network information according to the expected income and the offset parameter, improves the accuracy of the estimated quantity of the crowd due to the correction on the estimated quantity of the crowd, further calculates the more accurate exposure probability and the offset parameter of the obtained network information, orders the network information according to the expected income and the offset parameter, maximizes the exposure effect of the network information, considers the estimated click rate of the network information for the exposure of each piece of network information, and selects the user with the highest click rate among potential users for the network information, the overall click rate of the network information is improved.
In addition, the directional type is adopted, the users can be finely distinguished, the network information distribution is improved from probability selection to competition selection, the user value is maximized, and the overall playing effect is improved.
In order to further explain the online ranking method of network information resources, the online ranking method of network information resources is applied to the online ranking of advertisement information for detailed description. The online ordering process of the network information resource online ordering method for the advertisement information comprises the following steps:
(1) and acquiring the estimated exposure and the actual exposure of the advertisement information.
(2) And obtaining the estimated error proportion under the orientation type according to the estimated exposure and the actual exposure.
(3) And expecting the estimated error proportion of the advertisement information under the orientation type through a time attenuation function to obtain the total estimated error proportion under the orientation type.
(3) And correcting the pre-acquired estimated crowd quantity according to the total estimated error proportion under the orientation type to obtain the corrected estimated crowd quantity under the orientation type.
(4) And calculating the exposure probability of the advertisement information under the directional type according to the corrected estimated population number.
Specifically, acquiring the estimated quantity of the targeted crowd meeting the advertising information; determining the priority of the advertisement information according to the estimated quantity of the satisfied targeted crowd; and calculating the exposure probability of the advertisement information under the targeting type from high to low according to the priority.
(5) Obtaining expected revenue and expected revenue distribution probability density of advertisement information exposure under the targeting type
Specifically, acquiring the statistical single click cost of the advertisement information and the estimated click rate of the advertisement information of each exposure; calculating the product of the single click cost and the estimated click rate to obtain the expected yield of each exposure of the advertisement information; and counting according to the expected revenue of each exposure of the advertisement information to obtain the expected revenue distribution probability density of the advertisement information.
(6) And counting the expected revenue distribution probability density of the non-contract advertisement information under the targeting type.
Specifically, the contract-less advertisement information refers to advertisement information other than contract-made.
(7) And calculating the bias parameters of the advertisement information under the targeting type through an accumulative probability function according to the exposure probability of the advertisement information under the targeting type, the expected revenue distribution probability density of the unconventional advertisement information and the expected revenue distribution probability density of each piece of advertisement information.
(8) And ordering the advertisement information on line according to the expected income and the bias parameters.
The online ordering method of network information resources is applied to the ordering of advertisement information for description, and the online ordering method of network information resources can also be applied to the online ordering of article information, news information, activity information and the like, and the processing processes are the same, and are not repeated herein.
Fig. 7 is a schematic structural diagram of an online ranking apparatus for network information resources in an embodiment. As shown in FIG. 7, an online ranking apparatus for network information resources comprises an exposure acquisition module 710, an estimated error ratio acquisition module 720, a correction module 730, an exposure probability calculation module 740, an expected profit acquisition module 750, a bias parameter acquisition module 760 and a ranking module 770. Wherein:
the exposure acquisition module 710 is used for acquiring the estimated exposure and the actual exposure of the network information.
In this embodiment, the exposure obtaining module 710 is further configured to count historical display times within a predetermined time under the directional type, and estimate the estimated exposure of the network information according to the historical display times.
Specifically, the targeting type refers to a category under a targeting crowd preset in the network information, such as 1 year old, 2 years old or 1 to 6 years old for a male. The predetermined time may be set as desired, such as 5 days, 7 days, or one month. The historical display times in the preset time can be used as the estimated exposure of the network information, or the historical display times in the preset time can be weighted and averaged to be used as the estimated exposure of the network information, and the like.
The estimated error ratio obtaining module 720 is used for obtaining the estimated error ratio under the orientation type according to the estimated exposure and the actual exposure.
Specifically, the estimated error ratio may be calculated once at predetermined time intervals, which may be 5 minutes, 10 minutes, and the like. And correcting the deviation of the estimated exposure under each orientation type according to the real-time exposure and the estimated exposure of the network information server and the orientation type of the network information each time. The specific correction formula is as follows (1):
in the formula (1), ujIndicating the network information j actual exposure, djIndicating the estimated exposure, σ, of the network information j during this timeitAnd (4) representing the exposure deviation of the orientation type i at the time t, namely the estimated error proportion under the orientation type.
The correcting module 730 is configured to correct the pre-acquired estimated population number according to the estimated error ratio in the orientation type to obtain the corrected estimated population number in the orientation type.
In this embodiment, the correcting module 730 is further configured to expect the estimated error proportion of each piece of network information in the orientation type through a time decay function, so as to obtain a total estimated error proportion in the orientation type; and correcting the pre-acquired estimated crowd quantity according to the total estimated error proportion under the orientation type to obtain the corrected estimated crowd quantity under the orientation type.
In particular, σi＝Ef(t)[σit] (2)
f(t)＝λe-λt (3)
In the formulas (2) and (3), f (t) is a time decay function which is expressed by exponential distribution, the value of lambda can be determined according to the actual situation, the strength of time decay is expressed, and the smaller the lambda is, the smaller the lambda isThe larger the dependence on the historical data is, in this embodiment, λ may be 0.5; t represents the time from the present and can range from 0 to 48 in units of half an hour. SigmaiAnd (3) representing the total estimated error proportion under the orientation type i, wherein the estimated error proportion in each time period is expected by using a time attenuation function.
According to the calculated total estimated error proportion sigmaiCorrecting the pre-acquired estimated population quantity, wherein the correction formula is as follows (4):
in formula (4), SupplyiRepresenting the pre-acquired estimate of the population,
indicating the estimated population after correction.
The exposure probability calculation module 740 is configured to calculate the exposure probability of the network information in the directional type according to the corrected estimated population quantity.
In this embodiment, the exposure probability calculation module 740 is further configured to obtain an estimated number of directional people that satisfy the network information; determining the priority of the network information according to the satisfied estimated quantity of the directional crowd; and calculating the exposure probability of the network information under the orientation type from high to low according to the priority.
For example, the number of people satisfied by the first network information is 100W (ten thousands of people) +80W, and 180W in total; the number of people satisfied by the second network information is 80W +120W, and the total number of people is 200W. The priority of the first network information is higher than the priority of the second network information.
The specific process of calculating the exposure probability of the network information is as follows:
1) initializing remaining supply r for all orientation types ii＝si；
2) For the network information j, sorting according to the priority, and traversing a) and b);
a) solving the following equation (5) to obtain alphaj；
If the solution is not available, then set alphaj＝1。
b) For all i e (j), update ri＝ri-min{ri,siαj}。
Due to riIndicating the remaining population exposure dose, which decreases each time a network message is assigned an exposure dose, so r is updated each iterationi。
The expected revenue acquisition module 750 is operable to acquire an expected revenue distribution probability density for the network information exposure at the type of orientation.
In this embodiment, the expected revenue obtaining module 750 is further configured to obtain the statistical single click rate of the network information and the estimated click rate of the network information for each exposure; calculating the product of the single click cost and the estimated click rate to obtain the expected yield of each exposure of the network information; and carrying out statistics according to the expected income of each exposure of the network information to obtain the expected income distribution probability density of the network information.
Specifically, when each piece of network information just starts to be on-line, the exposure probability of the network information is directly used for on-line selection, and the sequencing exposure is carried out according to the exposure probability of the network information.
After accumulating a predetermined number of exposures and clicks, a statistical CPC (cost per click) of the network information is calculated, denoted by stat _ CPC, where:
stat_cpc＝cost/click (6)
in the formula (6), cost represents the cost in the preset time of the network information, and click represents the number of clicks in the preset time. The predetermined number of exposures may be selected to be 10000 times and the number of clicks may be 100 times.
And calculating the ECPM (expected Cost Per future expressions, which is the expected income after 1000 network information exposures) of each network information exposure, wherein the calculation formula is shown as the formula (7).
ECPM＝stat_cpc*pCtr (7)
Wherein pCtr represents the estimated click rate of the network information of the current exposure.
The expected income of each exposure of each piece of network information is counted to obtain the expected income distribution probability density f of each piece of network informationcontract(Ecpm)。
The bias parameter obtaining module 760 is configured to obtain a bias parameter of the network information in the directional type according to the exposure probability of the network information in the directional type and the corresponding expected revenue distribution probability density.
In this embodiment, the bias parameter obtaining module 760 is further configured to count an expected revenue distribution probability density of the unconventional network information in the directional type; and calculating the bias parameters of the network information under the directional type through an accumulative probability function according to the exposure probability of the network information under the directional type, the expected profit distribution probability density of the unconventional network information and the expected profit distribution probability density of the network information.
Specifically, the probability density f of expected profit distribution of unconventional network information under the oriented type is countednon-contract(Ecpm) obtained by log statistics, such as the number of exposures of Ecpm at [ 0, 0.1, Ecpm at [ 0.1, 0.2, etc.
And (3) obtaining a bias parameter bias of the network information through an accumulative probability function F, and calculating a formula as the formula (8).
F(Ecpmcontract+Ecpmnon-contract+bias)＝α (8)
In the formula (8), α represents a network information exposure probability.
The ranking module 770 is configured to rank the network information online according to the expected revenue and the bias parameters.
Specifically, the network information is sorted by taking the expected profit and the bias parameter as sorting factors.
The network information resource online sequencing device obtains the estimated error proportion through the estimated exposure and the actual exposure of the network information, corrects the estimated quantity of people according to the estimated error proportion, obtains the exposure probability of the network information according to the estimated quantity of people, obtains the offset parameter of the network information according to the expected income distribution probability density and the exposure probability, and sequences the network information online according to the expected income and the offset parameter, so that the accuracy of the estimated quantity of people is improved due to the correction of the estimated quantity of people, the exposure probability and the offset parameter of the obtained network information are more accurate, and sequences the network information according to the expected income and the offset parameter, thereby maximizing the exposure effect of the network information, the estimated click rate of each piece of network information is considered, and the network information selects the user with the highest click rate among potential users, the overall click rate of the network information is improved.
It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above can be implemented by a computer program, which can be stored in a non-volatile computer-readable storage medium, and can include the processes of the embodiments of the methods described above when the program is executed. The storage medium may be a magnetic disk, an optical disk, a Read-Only Memory (ROM), or the like.
The above-mentioned embodiments only express several embodiments of the present invention, and the description thereof is more specific and detailed, but not construed as limiting the scope of the present invention. It should be noted that, for a person skilled in the art, several variations and modifications can be made without departing from the inventive concept, which falls within the scope of the present invention. Therefore, the protection scope of the present patent shall be subject to the appended claims.
Claims (18)
1. A network information resource online sequencing method comprises the following steps:
acquiring estimated exposure and actual exposure of network information;
obtaining an estimated error proportion under the orientation type according to the estimated exposure and the actual exposure;
correcting the pre-acquired estimated crowd quantity according to the estimated error proportion under the orientation type to obtain the corrected estimated crowd quantity under the orientation type;
calculating the exposure probability of the network information under the directional type according to the estimated number of the corrected crowd, wherein the exposure probability of the network information refers to the distribution proportion of the network information;
acquiring expected income and expected income distribution probability density of network information exposure under the orientation type;
acquiring a bias parameter of the network information under the orientation type according to the exposure probability of the network information under the orientation type and the corresponding expected revenue distribution probability density;
and carrying out online sequencing on the network information according to the expected income and the bias parameters.
2. The method of claim 1, wherein the step of deriving the corrected estimated population quantity for the orientation type by correcting the pre-acquired estimated population quantity for the orientation type based on the estimated error ratio for the orientation type comprises:
expecting the estimated error proportion of the network information under the orientation type through a time attenuation function to obtain the total estimated error proportion under the orientation type;
and correcting the pre-acquired estimated crowd quantity according to the total estimated error proportion under the orientation type to obtain the corrected estimated crowd quantity under the orientation type.
3. The method of claim 1, wherein the step of calculating the exposure probability of the network information in the directional type according to the estimated corrected population amount comprises:
acquiring the estimated quantity of the directional crowd meeting the network information;
determining the priority of the network information according to the satisfied estimated quantity of the directional crowd;
and calculating the exposure probability of the network information under the orientation type from high to low according to the priority.
4. The method of claim 1, wherein the step of obtaining the expected revenue and expected revenue distribution probability density of the network information exposure in the directed type comprises:
acquiring the statistical single click cost of the network information and the estimated click rate of the network information of each exposure;
calculating the product of the single click cost and the estimated click rate to obtain the expected yield of each exposure of the network information;
and carrying out statistics according to the expected income of each exposure of the network information to obtain the expected income distribution probability density of the network information.
5. The method of claim 4, wherein the step of obtaining the bias parameters of the network information under the directional type according to the exposure probability of the network information under the directional type and the corresponding expected revenue distribution probability density comprises:
counting expected revenue distribution probability density of the unconventional network information under the orientation type;
and calculating the bias parameters of the network information under the directional type through an accumulative probability function according to the exposure probability of the network information under the directional type, the expected income distribution probability density of the unconventional network information and the expected income distribution probability density of the network information.
6. The method of claim 5, wherein the step of counting the expected revenue distribution probability density of the non-agreed network information under the directive type comprises:
and counting the expected profit distribution probability density of the unconventional network information under the orientation type from the log.
7. The method of claim 1, wherein the step of obtaining the estimated exposure of the network information comprises:
counting historical display times within preset time under the orientation type;
and estimating the estimated exposure of the network information according to the historical display times.
8. The method according to any one of claims 1 to 7, wherein the network information is advertisement information or article information or news information.
9. An online network information resource sequencing device, comprising:
the exposure acquisition module is used for acquiring the estimated exposure and the actual exposure of the network information;
the estimated error proportion acquisition module is used for acquiring an estimated error proportion under the orientation type according to the estimated exposure and the actual exposure;
the correction module is used for correcting the pre-acquired estimated crowd quantity according to the estimated error proportion under the orientation type to obtain the corrected estimated crowd quantity under the orientation type;
the exposure probability calculation module is used for calculating the exposure probability of the network information under the directional type according to the corrected estimated population quantity, wherein the exposure probability of the network information refers to the distribution proportion of the network information;
an expected income acquisition module, configured to acquire an expected income and an expected income distribution probability density of the network information exposure in the orientation type;
the bias parameter acquisition module is used for acquiring bias parameters of the network information under the orientation type according to the exposure probability of the network information under the orientation type and the corresponding expected revenue distribution probability density;
and the sequencing module is used for carrying out online sequencing on the network information according to the expected income and the bias parameters.
10. The apparatus according to claim 9, wherein the correcting module is further configured to expect the estimated error ratio of the network information in the directional type through a time decay function to obtain a total estimated error ratio in the directional type, and correct the pre-obtained estimated population amount according to the total estimated error ratio in the directional type to obtain the corrected estimated population amount in the directional type.
11. The apparatus according to claim 9, wherein the exposure probability calculating module is further configured to obtain an estimated number of directional people who satisfy the network information, determine a priority of the network information according to the estimated number of directional people who satisfy, and calculate the exposure probability of the network information in the directional type sequentially from high to low according to the priority.
12. The apparatus according to claim 9, wherein the expected profit obtaining module is further configured to obtain a statistical single click cost of the network information and an estimated click rate of the network information for each exposure, calculate a product of the single click cost and the estimated click rate to obtain an expected profit for each exposure of the network information, and perform statistics according to the expected profit for each exposure of the network information to obtain an expected profit distribution probability density of the network information.
13. The apparatus of claim 12, wherein the bias parameter obtaining module is further configured to count an expected revenue distribution probability density of the non-agreed network information under the orientation type, and calculate the bias parameter of the network information under the orientation type through an accumulative probability function according to the exposure probability of the network information under the orientation type, the expected revenue distribution probability density of the non-agreed network information, and the expected revenue distribution probability density of the network information.
14. The apparatus of claim 13, wherein the bias parameter obtaining module is further configured to count a probability density of expected revenue distribution of the non-agreed network information under the orientation type from a log.
15. The apparatus of claim 9, wherein the exposure dose obtaining module is further configured to count historical exposure times within a predetermined time period in the orientation type, and estimate the estimated exposure dose of the network information according to the historical exposure times.
16. The apparatus according to any one of claims 9 to 15, wherein the network information is advertisement information or article information or news information.
17. A server comprising a memory and a processor, the memory storing a computer program, characterized in that the processor realizes the steps of the method of any one of claims 1 to 8 when executing the computer program.
18. A non-transitory computer-readable storage medium storing a computer program, wherein the computer program, when executed by a processor, implements the steps of the method of any one of claims 1 to 8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510276519.7A CN106296247B (en) | 2015-05-26 | 2015-05-26 | Network information resource online ordering method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510276519.7A CN106296247B (en) | 2015-05-26 | 2015-05-26 | Network information resource online ordering method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106296247A CN106296247A (en) | 2017-01-04 |
| CN106296247B true CN106296247B (en) | 2020-12-22 |
Family
ID=57635331
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510276519.7A Active CN106296247B (en) | 2015-05-26 | 2015-05-26 | Network information resource online ordering method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106296247B (en) |
Families Citing this family (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107105031A (en) * | 2017-04-20 | 2017-08-29 | 北京京东尚科信息技术有限公司 | Information-pushing method and device |
| CN108965936B (en) * | 2017-05-17 | 2021-05-11 | 腾讯科技（深圳）有限公司 | Media information playing control method, server and computer storage medium |
| CN108959324B (en) * | 2017-05-26 | 2022-04-15 | 腾讯科技（深圳）有限公司 | Method and device for estimating multimedia display resource inventory and storage medium |
| CN109657132B (en) * | 2017-10-11 | 2023-04-25 | 腾讯科技（深圳）有限公司 | Recommended information cost control method, recommended information cost control device, computer equipment and storage medium |
| CN112150182B (en) * | 2019-06-28 | 2023-08-29 | 腾讯科技（深圳）有限公司 | Multimedia file pushing method and device, storage medium and electronic device |
| CN110599250B (en) * | 2019-09-09 | 2023-12-19 | 腾讯科技（深圳）有限公司 | Resource playing control method and device and computer storage medium |
| CN112581156A (en) * | 2019-09-30 | 2021-03-30 | 阿里巴巴集团控股有限公司 | Method, device, platform and medium for signing, flow distribution of creative script object |
| CN113516495B (en) * | 2020-09-30 | 2024-03-08 | 腾讯科技（深圳）有限公司 | Information pushing method, device, electronic equipment and computer readable medium |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102592235A (en) * | 2011-12-28 | 2012-07-18 | 北京品友互动信息技术有限公司 | Internet advertisement serving system |
| CN104182801A (en) * | 2013-05-22 | 2014-12-03 | 阿里巴巴集团控股有限公司 | Method and device for predicting website visits |
| CN104268644A (en) * | 2014-09-23 | 2015-01-07 | 新浪网技术(中国)有限公司 | Method and device for predicting click frequency of advertisement at advertising position |
| CN104424291A (en) * | 2013-09-02 | 2015-03-18 | 阿里巴巴集团控股有限公司 | Method and device for sorting search results |
| CN104572734A (en) * | 2013-10-23 | 2015-04-29 | 腾讯科技（深圳）有限公司 | Question recommendation method, device and system |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20140236710A1 (en) * | 2013-02-19 | 2014-08-21 | Congoo, Llc | On-line advertising valuation |
-
2015
- 2015-05-26 CN CN201510276519.7A patent/CN106296247B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102592235A (en) * | 2011-12-28 | 2012-07-18 | 北京品友互动信息技术有限公司 | Internet advertisement serving system |
| CN104182801A (en) * | 2013-05-22 | 2014-12-03 | 阿里巴巴集团控股有限公司 | Method and device for predicting website visits |
| CN104424291A (en) * | 2013-09-02 | 2015-03-18 | 阿里巴巴集团控股有限公司 | Method and device for sorting search results |
| CN104572734A (en) * | 2013-10-23 | 2015-04-29 | 腾讯科技（深圳）有限公司 | Question recommendation method, device and system |
| CN104268644A (en) * | 2014-09-23 | 2015-01-07 | 新浪网技术(中国)有限公司 | Method and device for predicting click frequency of advertisement at advertising position |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106296247A (en) | 2017-01-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106296247B (en) | Network information resource online ordering method and device | |
| Cai et al. | Real-time bidding by reinforcement learning in display advertising | |
| CN105447724B (en) | Content item recommendation method and device | |
| US7908238B1 (en) | Prediction engines using probability tree and computing node probabilities for the probability tree | |
| JP6522160B2 (en) | INFORMATION DELIVERY METHOD, APPARATUS, SERVER, AND STORAGE MEDIUM | |
| CN101673385A (en) | Consumption preliminary estimate method and device thereof | |
| CN109615442B (en) | RTB real-time bidding method based on incentive video advertisement | |
| CN111798280B (en) | Multimedia information recommendation method, device and equipment and storage medium | |
| Geyik et al. | Multi-touch attribution based budget allocation in online advertising | |
| US20140278937A1 (en) | Systems and Methods for Predicting and Pricing of Gross Rating Point Scores by Modeling Viewer Data | |
| US20120078730A1 (en) | Automatic Internet Search Advertising Campaign Variable Optimization for Aiding Advertising Agency Efficiencies | |
| US10181130B2 (en) | Real-time updates to digital marketing forecast models | |
| CN108564404B (en) | Method and device for predicting return on investment of advertisement | |
| JP6946082B2 (en) | Advertisement distribution support device, advertisement distribution support method, and program | |
| US20220027959A1 (en) | Real-Time Bidding | |
| CN108932627B (en) | Estimation method and device for consumption speed of CPA (continuous advertisement access) advertisement and readable storage medium | |
| US20130246161A1 (en) | Systems and methods for optimization-aware delivery pacing adjustment in advertisement serving | |
| Choi et al. | Optimizing reserve prices in display advertising auctions | |
| CN113159854A (en) | Exposure data determination method, device, equipment and storage medium | |
| CN113487367B (en) | Advertisement position prediction system based on cloud computing | |
| CN115545734A (en) | Recommendation information processing method, recommendation information sorting method and device | |
| CN115204922A (en) | Data prediction method and device of promotion content, electronic equipment and storage medium | |
| CN111899041B (en) | Information delivery processing and delivery method, device, equipment and storage medium | |
| CN110555709A (en) | method, device and storage medium for simulating advertising effect | |
| CN111899041A (en) | Information delivery processing method, information delivery device, information delivery equipment and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |