CN107862556A - A kind of put-on method and system of VIP advertisements - Google Patents
A kind of put-on method and system of VIP advertisements Download PDFInfo
- Publication number
- CN107862556A CN107862556A CN201711261332.5A CN201711261332A CN107862556A CN 107862556 A CN107862556 A CN 107862556A CN 201711261332 A CN201711261332 A CN 201711261332A CN 107862556 A CN107862556 A CN 107862556A
- Authority
- CN
- China
- Prior art keywords
- user
- vip
- vip user
- positive sample
- training
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0241—Advertisements
- G06Q30/0242—Determining effectiveness of advertisements
- G06Q30/0244—Optimization
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0241—Advertisements
- G06Q30/0242—Determining effectiveness of advertisements
- G06Q30/0243—Comparative campaigns
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0241—Advertisements
- G06Q30/0251—Targeted advertisements
- G06Q30/0255—Targeted advertisements based on user history
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0241—Advertisements
- G06Q30/0251—Targeted advertisements
- G06Q30/0257—User requested
Landscapes
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Accounting & Taxation (AREA)
- Development Economics (AREA)
- Strategic Management (AREA)
- Finance (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Economics (AREA)
- Marketing (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention provides a kind of put-on method of VIP advertisements, including：When the dispensing for receiving VIP advertisements is asked, the behavior label of each non-VIP user in the first preset duration is obtained；The behavior label of each non-VIP user is passed into default VIP user in predicting model as input quantity and carries out probable value calculating, obtaining each non-VIP user's registrations turns into the probability value set of VIP user；Each probable value in the probability value set is compared with default dispensing probability threshold value respectively, probable value is more than the non-VIP user of the dispensing probability threshold value as non-VIP user to be put；Give the VIP advertisement puttings to the non-VIP user to be put.Above-mentioned put-on method, determine that each non-VIP user's registrations turn into the probability of VIP user, VIP advertisement puttings are more than to the non-VIP user for launching probability threshold value to probability, dispensing is that targetedly, the possibility that each non-VIP user's registration turns into VIP user improves.
Description
Technical field
The present invention relates to advertisement putting field, more particularly to a kind of put-on method and system of VIP advertisements.
Background technology
Nowadays, some video websites carry out extra earning, for example, passing through to realize the maximization of interests using various means
More user's registrations are encouraged to increase the space of profit as the mode of video website payment VIP member users, in order to make more
More users turns into VIP member users, it is necessary to be promoted to VIP business, and the Main Means of popularization are by video website
User launch VIP advertisements and increase the registration amount of VIP member users.
Research discovery is carried out to the dispensing strategy of existing VIP advertisements, in VIP advertisement puttings by the way of logical throwing,
VIP advertisements can be pushed to each user.Above-mentioned put-on method, used because whether each video user registers VIP
The possibility at family is had differences, and the wasting of resources is easily caused by the way of logical throwing, and rate of return on investment is low.
The content of the invention
In view of this, the invention provides a kind of put-on method of VIP advertisements, to solve to throw using logical in the prior art
Mode easily cause the wasting of resources, the problem of rate of return on investment is low.Concrete scheme is as follows：
A kind of put-on method of VIP advertisements, including：
When the dispensing for receiving VIP advertisements is asked, the behavior mark of each non-VIP user in the first preset duration is obtained
Label；
The behavior label of each non-VIP user is passed into default VIP user in predicting model as input quantity to enter
Row probable value calculates, and obtaining each non-VIP user's registrations turns into the probability value set of VIP user；
Each probable value in the probability value set is compared with default dispensing probability threshold value respectively, will be general
Rate value is more than the non-VIP user of the dispensing probability threshold value as non-VIP user to be put；
Give the VIP advertisement puttings to the non-VIP user to be put.
Above-mentioned method, it is preferred that when the dispensing for receiving VIP advertisements is asked, obtain each in the first preset duration
The behavior label of individual non-VIP user includes：
Obtain the user behaviors log and attribute information of each non-VIP user；
User behaviors log and attribute information based on each non-VIP user write forecast model Feature Engineering；
According to the aspect of model engineering of the prediction, the behavior label of acquisition each non-VIP user.
Above-mentioned method, it is preferred that characterized in that, using the behavior label of each non-VIP user as input quantity
Pass to default VIP user in predicting model and carry out probable value calculating, obtain each non-VIP user's registrations and used as VIP
The probability value set at family includes：
According to the behavior label of user, default VIP user in predicting model is determined；
According to default training rules, the VIP user in predicting model is trained；
When the training is completed, the behavior label of each non-VIP user is delivered to the VIP user as input quantity
Forecast model carries out probable value calculating, obtains the probability value set for being registered as VIP user of each non-VIP user.
Above-mentioned method, it is preferred that according to default training rules, be trained, wrap to the VIP user in predicting model
Include：
Rule is determined according to default, determines that positive sample user collects；
Positive sample user collection is split, obtains positive sample user training set and positive sample user test collection；
The behavior label pair of each user is concentrated according to the positive sample user training set and the positive sample user test
The training and checking of the VIP user in predicting model；
When training result and the result, which are satisfied by corresponding prediction, to be required, training is completed.
Above-mentioned method, it is preferred that determine rule according to default, determine that positive sample user collection includes：
Filter out each non-VIP user to be put in the first preset duration；
Each non-VIP user to be put is registered to VIP user in the second preset duration, is designated as positive sample use
Family collection；
First preset duration has the time of identical start time and the second preset duration length with the second preset duration
Spend the time span for the first preset duration twice.
Above-mentioned method, it is preferred that concentrated according to the positive sample user training set and the positive sample user test each
Training and checking of the behavior label of individual user to the VIP user in predicting model include：
Obtain the positive sample user training set and the behavior label of each user is concentrated in the positive sample user test
Behavior label according to each positive sample user in positive sample user's training set is pre- to the default VIP user
Model is surveyed to be trained；
When the training is completed, the behavior label of each positive sample user is concentrated to described pre- according to positive sample user test
If VIP user in predicting models verified.
Above-mentioned method, it is preferred that also include：
Each non-VIP user in first preset duration is screened.
A kind of jettison system of VIP advertisements, including：
Acquisition module, for when the dispensing for receiving VIP advertisements is asked, obtaining each non-in the first preset duration
The behavior label of VIP user；
Computing module, used for the behavior label of each non-VIP user to be passed into default VIP as input quantity
Family forecast model carries out probable value calculating, and obtaining each non-VIP user's registrations turns into the probability value set of VIP user；
Contrast module, for by each probable value in the probability value set respectively with default dispensing probability threshold value
It is compared, probable value is more than the non-VIP user of the dispensing probability threshold value as non-VIP user to be put；
Putting module, for giving the VIP advertisement puttings to the non-VIP user to be put.
Above-mentioned system, it is preferred that the computing module includes：
Determining unit, for the behavior label according to user, determine default VIP user in predicting model；Training unit, use
According to default training rules, the VIP user in predicting model is trained；
Computing unit, for when the training is completed, using the behavior label of each non-VIP user as input quantity transmission
Probable value calculating is carried out to the VIP user in predicting model, obtains each non-VIP user's registrations as the general of VIP user
Rate value set.
Above-mentioned system, it is preferred that the training unit includes：
Determination subelement, for determining rule according to default, determine that positive sample user collects；
Subelement is split, positive sample user collection is split, positive sample user training set is obtained and positive sample is used
Family test set；
Training checking subelement, each use is concentrated according to the positive sample user training set and the positive sample user test
Training and checking of the behavior label at family to the VIP user in predicting model；
Subelement is completed, for when prediction corresponding to training result and the result are satisfied by requires, completing training.
Compared with prior art, the present invention includes advantages below：
The invention provides a kind of put-on method of VIP advertisements, including：When the dispensing for receiving VIP advertisements is asked, obtain
Take the behavior label of each non-VIP user in the first preset duration；Using the behavior label of each non-VIP user as defeated
Enter amount and pass to default VIP user in predicting model progress probable value calculating, obtaining each non-VIP user's registrations turns into
The probability value set of VIP user；By each probable value in the probability value set respectively with default dispensing probability threshold value
It is compared, probable value is more than the non-VIP user of the dispensing probability threshold value as non-VIP user to be put；By described in
The non-VIP user to be put is given in VIP advertisement puttings.Above-mentioned put-on method, determine that each non-VIP user's registrations turn into
The probability of VIP user, VIP advertisement puttings are more than to the non-VIP user for launching probability threshold value to probability, dispensing is targeted
, the possibility that each non-VIP user's registration turns into VIP user improves.
Brief description of the drawings
In order to illustrate more clearly about the embodiment of the present invention or technical scheme of the prior art, below will be to embodiment or existing
There is the required accompanying drawing used in technology description to be briefly described, it should be apparent that, drawings in the following description are only this
Some embodiments of invention, for those of ordinary skill in the art, on the premise of not paying creative work, can be with
Other accompanying drawings are obtained according to these accompanying drawings.
Fig. 1 is a kind of put-on method flow chart of VIP advertisements disclosed in the embodiment of the present application；
Fig. 2 is a kind of put-on method another method flow chart of VIP advertisements disclosed in the embodiment of the present application；
Fig. 3 is a kind of another method flow diagram of put-on method of VIP advertisements disclosed in the embodiment of the present application；
Fig. 4 is a kind of put-on method another method flow chart of VIP advertisements disclosed in the embodiment of the present application；
Fig. 5 is a kind of jettison system structured flowchart of VIP advertisements disclosed in the embodiment of the present application；
Fig. 6 is a kind of another structured flowchart of dispensing of VIP advertisements disclosed in the embodiment of the present application.
Embodiment
Below in conjunction with the accompanying drawing in the embodiment of the present invention, the technical scheme in the embodiment of the present invention is carried out clear, complete
Site preparation describes, it is clear that described embodiment is only part of the embodiment of the present invention, rather than whole embodiments.It is based on
Embodiment in the present invention, those of ordinary skill in the art are obtained every other under the premise of creative work is not made
Embodiment, belong to the scope of protection of the invention.
The foregoing description of the disclosed embodiments, professional and technical personnel in the field are enable to realize or using the present invention.
A variety of modifications to these embodiments will be apparent for those skilled in the art, as defined herein
General Principle can be realized in other embodiments without departing from the spirit or scope of the present invention.Therefore, it is of the invention
The embodiments shown herein is not intended to be limited to, and is to fit to and principles disclosed herein and features of novelty phase one
The most wide scope caused.
The invention provides a kind of put-on method of VIP advertisements, methods described is applied to website and carries out VIP advertisement puttings
During, the executive agent of the put-on method can be processor, release platform or putting module etc..The put-on method
Flow as shown in figure 1, including step：
S101, when the dispensing for receiving VIP advertisements is asked, obtain each non-VIP user in the first preset duration
Behavior label；
In the embodiment of the present invention, first preset duration can be to grow the regular time such as one week, one day or one hour
Degree.
S102, the behavior label of each non-VIP user passed into default VIP user in predicting mould as input quantity
Type carries out probable value calculating, and obtaining each non-VIP user's registrations turns into the probability value set of VIP user；
In the embodiment of the present invention, each non-VIP user corresponds to a probable value collection for being registered as VIP user.
S103, by each probable value in the probability value set respectively with it is default dispensing probability threshold value compared
It is right, probable value is more than the non-VIP user of the dispensing probability threshold value as non-VIP user to be put；
In the embodiment of the present invention, it is described launch probability threshold value selection can be empirical value can also be will be described each non-
VIP user's registrations choose a probable value as foundation actual conditions in the probability value set of VIP user.
S104, give the VIP advertisement puttings to the non-VIP user to be put.
In the embodiment of the present invention, the behavior label includes behavioural information and attribute information, and the behavioural information includes：Happiness
Joyous star, the channel for liking seeing, the video for liking seeing, active city, the type of equipment, the viewing amount of nearest one month,
The viewing number of days of nearest one month, the viewing amount of nearest one week, the viewing number of days of nearest one week, the VIP advertisements of nearest one week
Light exposure, the click volume of nearest VIP advertisements in one week, buy within nearest three months the number of VIP month cards, nearest three months purchase VIP
Season card number, nearest three months purchase VIP cards number, whether bought time gap today of VIP and last time viewing
Number of days etc.；The attribute information includes：Sex, age, education degree, occupation and income etc..
When needing to launch VIP advertisements, the user for being registered as VIP is not belonging to the focus of the present invention, it is only necessary to
The behavior label of each non-VIP user in preset duration is obtained, the first preset duration can be one week, one day or one hour
Deng.The first preset duration in the present invention was illustrated exemplified by one day, by the behavior label of each non-VIP user using pre-
If algorithm calculated, obtain the probability for being registered as VIP user of each non-VIP user, probability be more than pre-
If dispensing threshold value user of the non-VIP user as VIP advertisements to be put, the VIP advertisement puttings are waited to throw to described
The user for the VIP advertisements put.
Preferably, by the probability for being registered as VIP user of each non-VIP user, store to a VIP probability set
In.By it is described dispensing threshold value set rule as：Probability in the VIP probability sets is arranged according to descending, foundation is treated
The actual conditions of advertisement video website are launched, select a profit to be determined with launching the equalization point of ratio in the VIP probability sets
The dispensing ratio of corresponding non-VIP user, according to the dispensing ratio, search in the VIP probability sets with the dispensing ratio
The probable value of corresponding points, the probable value is defined as to launch probability threshold value.
The invention provides a kind of put-on method of VIP advertisements, including：When the dispensing for receiving VIP advertisements is asked, obtain
Take the behavior label of each non-VIP user in the first preset duration；Using the behavior label of each non-VIP user as defeated
Enter amount and pass to default VIP user in predicting model progress probable value calculating, obtaining each non-VIP user's registrations turns into
The probability value set of VIP user；By each probable value in the probability value set respectively with default dispensing probability threshold value
It is compared, probable value is more than the non-VIP user of the dispensing probability threshold value as non-VIP user to be put；By described in
The non-VIP user to be put is given in VIP advertisement puttings.Above-mentioned put-on method, determine that each non-VIP user's registrations turn into
The probability of VIP user, VIP advertisement puttings are more than to the non-VIP user for launching probability threshold value to probability, dispensing is targeted
, the possibility that each non-VIP user's registration turns into VIP user improves.
When the dispensing for receiving VIP advertisements is asked, the behavior label of each non-VIP user in acquisition preset duration
Method flow is as shown in Fig. 2 including step：
S201, the user behaviors log and attribute information for obtaining each non-VIP user；
In the embodiment of the present invention, the user behaviors log and attribute information of each non-VIP user are obtained, wherein the behavior day
Will includes the non-VIP user all behavioural informations in preset duration；The attribute information includes：Sex, age, education journey
Degree, occupation and income etc..
S202, the user behaviors log based on each non-VIP user and attribute information write forecast model Feature Engineering；
In the embodiment of the present invention, user behaviors log and attribute information based on each non-VIP user, a prediction mould is write
Type Feature Engineering, the aspect of model engineering of the prediction refers to extracts feature for algorithm and mould from initial data to greatest extent
Type uses, and the forecast model Feature Engineering can be one section of program.The development environment of aspect of model engineering in the present invention is
Pig, wherein Pig are that yahoo contributes a project to apache, using SQL-like language, are built on MapReduce
A kind of high level query language.
S203, the aspect of model engineering according to the prediction, obtain the behavior label of each non-VIP user.
In the embodiment of the present invention, each non-VIP user can correspond to many behavior labels, appoint and take a non-VIP user
After aspect of model project treatment by the prediction, obtained result can be：User：A, sex：Man, age：20, city
City：Shanghai, the channel liked：Film etc. behavior label.
In the embodiment of the present invention, the row of each non-VIP user can be obtained according to the preset model Feature Engineering
For label, the probability according to each non-VIP user's registrations of the behavior tag computation as VIP user.
The behavior label of each non-VIP user is passed into default VIP user in predicting model as input quantity to enter
Row probable value calculates, and obtains the method flow such as Fig. 3 of each non-VIP user's registrations as the probability value set of VIP user
It is shown, including step：
S301, the behavior label according to user, determine default VIP user in predicting model；
In the embodiment of the present invention, the user can be that VIP user can also be non-VIP user.By the behavior mark of user
Label regard two classification problems as, promote the behavior label of facilitation to regard wherein turning into VIP user to the user's registration and playing
For positive sample label, the behavior label for playing inhibition is considered as negative sample behavior label, therefore, by the behavior of each user
Label is divided into positive sample behavior label and the class of negative sample behavior label two.Wherein, two classification self-explanatory characters' thing is divided by a certain property
The only variable of two class results.For example male and female, landlord and tenant, successfully with failure, pass and fail, life or death etc.
Deng.
Due to being two classification problems, therefore, default VIP user in predicting model is used as logistic regression grader mould
Type, the behavior label of each user are related to a weight to whether to be registered as the influence degree of VIP user be different
The problem of, the logistic regression sorter model takes into account the weight of each behavioural characteristic, improves the accurate of calculating
Property.The determination of weight is also to be analyzed by the big data of the history viewing record of user, obtains each behavior label
Possible weight.
In the embodiment of the present invention, it is preferred that based on LR/GBDT structures, development environment during the VIP user in predicting model
Based on Spark.Certainly, the model of the invention protected is fallen within based on the VIP user in predicting models that other development environments are developed
Enclose.
S302, according to default training rules, the VIP user in predicting model is trained；
S303, when the training is completed, it is delivered to using the behavior label of each non-VIP user as input quantity described
VIP user in predicting model carries out probable value calculating, and obtaining each non-VIP user's registrations turns into the probable value collection of VIP user
Close.
In the embodiment of the present invention, the default differentiation threshold value is to pass through Receiver operating curve
(Receiver Operating Characteristic Curve, abbreviation ROC curve) is come what is determined.
In the embodiment of the present invention, during the behavior label to each non-VIP user is classified, to specific
For some VIP advertisement, some specific behavior label is sentencing for positive sample behavior label or negative sample behavior label
Surely there can be certain error, positive sample behavior label can be determined as to negative sample behavior label, it is also possible to by negative sample behavior
Label is determined as positive sample behavioural characteristic, above-mentioned this erroneous judgement, training result will be produced during model training
Influence, assess influence index and be called Receiver operating curve (Receiver Operating Characteristic
Curve, abbreviation ROC curve), what Receiver operating curve characterized is exactly the discrimination for the behavioural characteristic for aligning negative sample.
In the embodiment of the present invention, preferably when the area of Receiver operating curve corresponding with above-mentioned training process
When AUC is more than 0.7, illustrate very high to the discrimination of positive sample behavior label and negative sample behavior label, it is believed that the VIP user
Parameter setting in forecast model meets to require, the VIP user in predicting models can be used for each non-VIP user's registration into
Estimated for the probability of VIP user.
It is that each non-VIP user determines that one may register by above-mentioned computational methods in the embodiment of the present invention
As the probability of VIP user.
In the embodiment of the present invention, according to default training rules, the method being trained to the VIP user in predicting model
Flow is as shown in figure 4, including step：
S401, according to it is default determine rule, determine positive sample user collect；
In the embodiment of the present invention, each non-VIP user to be put in the first preset duration is filtered out, will be described each
Non- VIP user to be put registers VIP user in the second preset duration, is designated as positive sample user collection.Wherein, first is default
It is the first preset duration that duration and the second preset duration, which have the time span of identical start time and the second preset duration,
Twice of time span.For example, if the first preset duration of the broadcasting of VIP advertisements is the whole day on the 6th of September in 2017, that
The user for being registered as VIP between 7 days in 6 days -2017 Septembers of September in 2017 is positive sample user collection, wherein, 2017
6 days -2017 Septembers of September 7 days are default second preset duration.
S402, positive sample user collection is split, obtain positive sample user training set and positive sample user test
Collection；
In the embodiment of the present invention, the positive sample user is collected and split according to certain primary contract, primary contract
It is determined according to concrete condition.In the embodiment of the present invention, it is preferred that 70% user for concentrating the positive sample user makees
For training set, remaining 30% user is as test set.Specific fractionation mode can have a variety of ways of realization：Random point
Group, period packet or other preferred packet modes by user's registration.
S403, the behavior mark for concentrating according to the positive sample user training set and the positive sample user test each user
Sign the training and checking to VIP user in predicting models；
In the embodiment of the present invention, the user that 70% positive sample user is concentrated is obtained in the training set as training set
The behavior label of each user, the VIP user in predicting model is trained.
S404, when prediction corresponding to training result and the result are satisfied by requires, complete training.
In the embodiment of the present invention, illustrated by taking the determination process of a simple forecast model as an example, can be used for sentencing
The sex of disconnected user, input as the length of user's hair, the sex for user is exported, for example, user a, the long 20cm of hair, sex
Female, user b, the long 3cm of hair, sex man.User can learn sex (output) and input (namely refers to behavior label, herein
Refer to hair lengths) relation, if a user c, the long 30cm of hair, sex is unknown, and the model can is according to dispensing
The behavior label of length, predict c sex.Above-mentioned example, simply a simple citing, it is generally the case that input and be
One group (thousands of behavioural characteristic) is formed.
During the present invention is implemented, it is preferred that after the VIP user in predicting model completes training, also need to be verified have
The verification method of body is：The user of positive sample user concentration 30% is selected as test set.Each in test set is obtained to test
The behavior label and selection result of user, the behavioural characteristic of each test user in test set is pre- by the VIP user
Model is surveyed to be calculated.
It is very high with the situation goodness of fit of reality when selection is registered as that probable value corresponding to VIP user is larger in test set
When, complete the checking of the VIP user in predicting model.
In the embodiment of the present invention, in addition to：When the dispensing for receiving VIP advertisements is asked, it is also necessary to when default to first
Each non-VIP user is screened in length, filters out abnormal non-VIP user.Preferably, the abnormal non-VIP user can
Think do not occur the user of viewing behavior in one week and daily advertisement exposure amount be more than 1000 user, carried out above-mentioned pre-
The quantity of each non-VIP user in preset duration is further reduced after processing, reduces the workload of subsequent treatment.
In the embodiment of the present invention, in addition to：The user of the VIP advertisements to be put is formed into a VIP advertisement to be put
User's collection, the user for counting the VIP advertisements to be put concentrates the registration VIP of each non-VIP user quantity to be united
Meter, the sample for being registered as VIP user is concentrated to calculate the positive sample user as positive sample the VIP user to be put
Quantity account for the percentage that the VIP user to be put concentrates sum, the put-on method of above-mentioned VIP advertisements is assessed.
Corresponding with above-mentioned put-on method, present invention also offers a kind of jettison system of VIP advertisements, the dispensing
The structured flowchart of system as shown in figure 5, including：
Acquisition module 501, computing module 502, contrast module 503 and putting module 504.
Wherein,
The acquisition module 501, for when the dispensing for receiving VIP advertisements is asked, obtaining in the first preset duration
The behavior label of each non-VIP user；
The computing module 502, it is default for being passed to using the behavior label of each non-VIP user as input quantity
VIP user in predicting model carry out probable value calculating, obtain each non-VIP user's registrations turn into VIP user probable value
Set；
The contrast module 503, for by each probable value in the probability value set respectively with default dispensing
Probability threshold value is compared, and probable value is more than to use of the non-VIP user as VIP advertisements to be put of the dispensing probability threshold value
Family；
504 putting module, for the VIP advertisement puttings to be given to the user of the VIP advertisements to be put.
The invention provides a kind of jettison system of VIP advertisements, including：When the dispensing for receiving VIP advertisements is asked, obtain
Take the behavior label of each non-VIP user in the first preset duration；Using the behavior label of each non-VIP user as defeated
Enter amount and pass to default VIP user in predicting model progress probable value calculating, obtaining each non-VIP user's registrations turns into
The probability value set of VIP user；By each probable value in the probability value set respectively with default dispensing probability threshold value
It is compared, probable value is more than the non-VIP user of the dispensing probability threshold value as non-VIP user to be put；By described in
The non-VIP user to be put is given in VIP advertisement puttings.Above-mentioned jettison system, determine that each non-VIP user's registrations turn into
The probability of VIP user, VIP advertisement puttings are more than to the non-VIP user for launching probability threshold value to probability, dispensing is targeted
, the possibility that each non-VIP user's registration turns into VIP user improves.
In the embodiment of the present invention, the structured flowchart of the computing module 502 as shown in fig. 6, including：
Determining unit 505, training unit 506 and computing unit 507.
Wherein,
The determining unit 505, for the behavior label according to user, determine default VIP user in predicting model；
The training unit 506, for according to default training rules, being trained to the VIP user in predicting model；
The computing unit 507, for when the training is completed, using the behavior label of each non-VIP user as defeated
Enter amount and be delivered to the VIP user in predicting model progress probable value calculating, obtaining each non-VIP user's registrations turns into VIP
The probability value set of user.
In the embodiment of the present invention, the structured flowchart of the training unit 506 as shown in fig. 6, including：
Determination subelement 508, split subelement 509, training subelement 510 and complete subelement 511.
Wherein,
The determination subelement 508, for determining rule according to default, determine that positive sample user collects；
The fractionation subelement 509, splits to positive sample user collection, obtains positive sample user training set and just
Sample of users test set；
The training checking subelement 510, according to the positive sample user training set and the positive sample user test collection
In each user training and checking of the behavior label to the VIP user in predicting model.
The completion subelement 511, for when prediction corresponding to training result and the result are satisfied by requires, completing
Training.
It should be noted that each embodiment in this specification is described by the way of progressive, each embodiment weight
Point explanation is all difference with other embodiment, between each embodiment identical similar part mutually referring to.
For device class embodiment, because it is substantially similar to embodiment of the method, so description is fairly simple, related part is joined
See the part explanation of embodiment of the method.
Finally, it is to be noted that, in this application, such as first and second or the like relational terms are used merely to
One entity or operation are made a distinction with another entity or operation, and not necessarily require or imply these entities or behaviour
Any this actual relation or order between work be present.Moreover, term " comprising ", "comprising" or its any other variant
Including for nonexcludability is intended to, so that process, method, article or equipment including a series of elements not only include
Those key elements, but also the other element including being not expressly set out, or also include for this process, method, article or
The intrinsic key element of person's equipment.In the absence of more restrictions, the key element limited by sentence "including a ...", not
Other identical element in the process including the key element, method, article or equipment also be present in exclusion.
For convenience of description, it is divided into various units during description apparatus above with function to describe respectively.Certainly, this is being implemented
The function of each unit can be realized in same or multiple softwares and/or hardware during invention.
As seen through the above description of the embodiments, those skilled in the art can be understood that the present invention can
Realized by the mode of software plus required general hardware platform.Based on such understanding, technical scheme essence
On the part that is contributed in other words to prior art can be embodied in the form of software product, the computer software product
It can be stored in storage medium, such as ROM/RAM, magnetic disc, CD, including some instructions are causing a computer equipment
(can be personal computer, server, either network equipment etc.) performs some of each embodiment of the present invention or embodiment
Method described in part.
The put-on method and system of a kind of VIP advertisements provided by the present invention are described in detail above, herein
Apply specific embodiment to be set forth the principle and embodiment of the present invention, the explanation of above example is only intended to
Help to understand method and its core concept of the invention；Meanwhile for those of ordinary skill in the art, the think of according to the present invention
Think, in specific embodiments and applications there will be changes, in summary, this specification content should not be construed as pair
The limitation of the present invention.
Described above is only the preferred embodiment of the present invention, it is noted that for the ordinary skill people of the art
For member, under the premise without departing from the principles of the invention, some improvements and modifications can also be made, these improvements and modifications also should
It is considered as protection scope of the present invention.
Claims (10)
- A kind of 1. put-on method of VIP advertisements, it is characterised in that including：When the dispensing for receiving VIP advertisements is asked, the behavior label of each non-VIP user in the first preset duration is obtained；The behavior label of each non-VIP user is passed into default VIP user in predicting model as input quantity to carry out generally Rate value calculates, and obtaining each non-VIP user's registrations turns into the probability value set of VIP user；Each probable value in the probability value set is compared with default dispensing probability threshold value respectively, by probable value More than the non-VIP user for launching probability threshold value as non-VIP user to be put；Give the VIP advertisement puttings to the non-VIP user to be put.
- 2. according to the method for claim 1, it is characterised in that when the dispensing for receiving VIP advertisements is asked, obtain first The behavior label of each non-VIP user in preset duration includes：Obtain the user behaviors log and attribute information of each non-VIP user；User behaviors log and attribute information based on each non-VIP user write forecast model Feature Engineering；According to the aspect of model engineering of the prediction, the behavior label of acquisition each non-VIP user.
- 3. according to the method for claim 1, it is characterised in that using the behavior label of each non-VIP user as defeated Enter amount and pass to default VIP user in predicting model progress probable value calculating, obtaining each non-VIP user's registrations turns into The probability value set of VIP user includes：According to the behavior label of user, default VIP user in predicting model is determined；According to default training rules, the VIP user in predicting model is trained；When the training is completed, the behavior label of each non-VIP user is delivered to the VIP user in predicting as input quantity Model carries out probable value calculating, obtains the probability value set for being registered as VIP user of each non-VIP user.
- 4. according to the method for claim 3, it is characterised in that according to default training rules, to the VIP user in predicting Model is trained, including：Rule is determined according to default, determines that positive sample user collects；Positive sample user collection is split, obtains positive sample user training set and positive sample user test collection；The behavior label of each user is concentrated to described according to the positive sample user training set and the positive sample user test The training and checking of VIP user in predicting models；When training result and the result, which are satisfied by corresponding prediction, to be required, training is completed.
- 5. according to the method for claim 4, it is characterised in that determine rule according to default, determine that positive sample user collects Including：Filter out each non-VIP user to be put in the first preset duration；Each non-VIP user to be put is registered to VIP user in the second preset duration, is designated as positive sample user Collection；There is the time span of identical start time and the second preset duration to be for first preset duration and the second preset duration Twice of the time span of first preset duration.
- 6. according to the method for claim 4, it is characterised in that according to the positive sample user training set and the positive sample User test concentrates training and checking of the behavior label of each user to the VIP user in predicting model to include：Obtain the positive sample user training set and the behavior label of each user is concentrated in the positive sample user testBehavior label according to each positive sample user in positive sample user's training set is to the default VIP user in predicting mould Type is trained；When the training is completed, the behavior label of each positive sample user is concentrated to described default according to positive sample user test VIP user in predicting models are verified.
- 7. according to the method for claim 1, it is characterised in that also include：Each non-VIP user in first preset duration is screened.
- A kind of 8. jettison system of VIP advertisements, it is characterised in that including：Acquisition module, for when the dispensing for receiving VIP advertisements is asked, each non-VIP obtained in the first preset duration to be used The behavior label at family；Computing module, it is pre- for passing to default VIP user using the behavior label of each non-VIP user as input quantity Survey model and carry out probable value calculating, obtaining each non-VIP user's registrations turns into the probability value set of VIP user；Contrast module, for each probable value in the probability value set to be carried out with default dispensing probability threshold value respectively Compare, probable value is more than the non-VIP user of the dispensing probability threshold value as non-VIP user to be put；Putting module, for giving the VIP advertisement puttings to the non-VIP user to be put.
- 9. system according to claim 1, it is characterised in that the computing module includes：Determining unit, for the behavior label according to user, determine default VIP user in predicting model；Training unit, for according to According to default training rules, the VIP user in predicting model is trained；Computing unit, for when the training is completed, the behavior label of each non-VIP user being delivered into institute as input quantity State VIP user in predicting model and carry out probable value calculating, obtaining each non-VIP user's registrations turns into the probable value of VIP user Set.
- 10. system according to claim 9, it is characterised in that the training unit includes：Determination subelement, for determining rule according to default, determine that positive sample user collects；Subelement is split, positive sample user collection is split, positive sample user training set is obtained and positive sample user surveys Examination collection；Training checking subelement, concentrates each user's according to the positive sample user training set and the positive sample user test Training and checking of the behavior label to the VIP user in predicting model；Subelement is completed, for when prediction corresponding to training result and the result are satisfied by requires, completing training.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711261332.5A CN107862556A (en) | 2017-12-04 | 2017-12-04 | A kind of put-on method and system of VIP advertisements |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711261332.5A CN107862556A (en) | 2017-12-04 | 2017-12-04 | A kind of put-on method and system of VIP advertisements |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107862556A true CN107862556A (en) | 2018-03-30 |
Family
ID=61704542
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711261332.5A Pending CN107862556A (en) | 2017-12-04 | 2017-12-04 | A kind of put-on method and system of VIP advertisements |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107862556A (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110992127A (en) * | 2019-11-14 | 2020-04-10 | 北京沃东天骏信息技术有限公司 | Article recommendation method and device |
| CN111260383A (en) * | 2018-11-30 | 2020-06-09 | 北京嘀嘀无限科技发展有限公司 | Registration probability estimation method and device and probability estimation model construction method and device |
| CN111262716A (en) * | 2018-11-30 | 2020-06-09 | 北京嘀嘀无限科技发展有限公司 | Registration probability estimation method and device and probability estimation model construction method and device |
| CN112348587A (en) * | 2020-11-16 | 2021-02-09 | 脸萌有限公司 | Information pushing method and device and electronic equipment |
| CN112785344A (en) * | 2021-02-01 | 2021-05-11 | 北京达佳互联信息技术有限公司 | Advertisement putting method and device, electronic equipment and storage medium |
| CN112819536A (en) * | 2021-02-01 | 2021-05-18 | 北京奇艺世纪科技有限公司 | Effect advertisement display method and device, computer equipment and storage medium |
| CN116934389A (en) * | 2023-06-12 | 2023-10-24 | 嘉兴华数广电网络有限公司 | Digital television value ticket card management system based on cloud computing and cross-platform technology |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20090182621A1 (en) * | 2008-01-14 | 2009-07-16 | Dream Makers Music, Llc | Content and advertising material superdistribution |
| CN101493914A (en) * | 2008-01-24 | 2009-07-29 | 简子峻 | Marketing method for emulation e-book |
| CN105389713A (en) * | 2015-10-15 | 2016-03-09 | 南京大学 | Mobile data traffic package recommendation algorithm based on user historical data |
| CN105488697A (en) * | 2015-12-09 | 2016-04-13 | 焦点科技股份有限公司 | Potential customer mining method based on customer behavior characteristics |
| CN106204063A (en) * | 2016-06-30 | 2016-12-07 | 北京奇艺世纪科技有限公司 | A kind of paying customer's method for digging and device |
-
2017
- 2017-12-04 CN CN201711261332.5A patent/CN107862556A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20090182621A1 (en) * | 2008-01-14 | 2009-07-16 | Dream Makers Music, Llc | Content and advertising material superdistribution |
| CN101493914A (en) * | 2008-01-24 | 2009-07-29 | 简子峻 | Marketing method for emulation e-book |
| CN105389713A (en) * | 2015-10-15 | 2016-03-09 | 南京大学 | Mobile data traffic package recommendation algorithm based on user historical data |
| CN105488697A (en) * | 2015-12-09 | 2016-04-13 | 焦点科技股份有限公司 | Potential customer mining method based on customer behavior characteristics |
| CN106204063A (en) * | 2016-06-30 | 2016-12-07 | 北京奇艺世纪科技有限公司 | A kind of paying customer's method for digging and device |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111260383A (en) * | 2018-11-30 | 2020-06-09 | 北京嘀嘀无限科技发展有限公司 | Registration probability estimation method and device and probability estimation model construction method and device |
| CN111262716A (en) * | 2018-11-30 | 2020-06-09 | 北京嘀嘀无限科技发展有限公司 | Registration probability estimation method and device and probability estimation model construction method and device |
| CN111260383B (en) * | 2018-11-30 | 2023-08-29 | 北京嘀嘀无限科技发展有限公司 | Registration probability estimation method and device and probability estimation model construction method and device |
| CN110992127A (en) * | 2019-11-14 | 2020-04-10 | 北京沃东天骏信息技术有限公司 | Article recommendation method and device |
| CN110992127B (en) * | 2019-11-14 | 2023-09-29 | 北京沃东天骏信息技术有限公司 | Article recommendation method and device |
| CN112348587A (en) * | 2020-11-16 | 2021-02-09 | 脸萌有限公司 | Information pushing method and device and electronic equipment |
| CN112348587B (en) * | 2020-11-16 | 2024-04-23 | 脸萌有限公司 | Information pushing method and device and electronic equipment |
| CN112785344A (en) * | 2021-02-01 | 2021-05-11 | 北京达佳互联信息技术有限公司 | Advertisement putting method and device, electronic equipment and storage medium |
| CN112819536A (en) * | 2021-02-01 | 2021-05-18 | 北京奇艺世纪科技有限公司 | Effect advertisement display method and device, computer equipment and storage medium |
| CN112819536B (en) * | 2021-02-01 | 2023-09-01 | 北京奇艺世纪科技有限公司 | Method, device, computer equipment and storage medium for displaying effect advertisement |
| CN116934389A (en) * | 2023-06-12 | 2023-10-24 | 嘉兴华数广电网络有限公司 | Digital television value ticket card management system based on cloud computing and cross-platform technology |
| CN116934389B (en) * | 2023-06-12 | 2024-02-23 | 嘉兴华数广电网络有限公司 | Digital television value ticket card management system based on cloud computing and cross-platform technology |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107862556A (en) | A kind of put-on method and system of VIP advertisements | |
| CN110612525B (en) | Enabling a tutorial analysis by using an alternating speech tree | |
| Yu et al. | k-Nearest neighbor model for multiple-time-step prediction of short-term traffic condition | |
| US10235683B2 (en) | Analyzing mobile-device location histories to characterize consumer behavior | |
| US9245252B2 (en) | Method and system for determining on-line influence in social media | |
| JP2019527874A (en) | Predict psychometric profiles from behavioral data using machine learning while maintaining user anonymity | |
| Xie et al. | AppWatcher: Unveiling the underground market of trading mobile app reviews | |
| Hubbard | Pulse: The new science of harnessing internet buzz to track threats and opportunities | |
| Christoforidis et al. | Recommendation of points-of-interest using graph embeddings | |
| CN111126495B (en) | Model training method, information prediction device, storage medium and equipment | |
| Ottomanelli et al. | Modelling parking choice behaviour using Possibility Theory | |
| CN111061979B (en) | User tag pushing method and device, electronic equipment and medium | |
| WO2014193700A1 (en) | Social media pricing engine | |
| Pang et al. | Adaptive recommendation for MOOC with collaborative filtering and time series | |
| US9342787B2 (en) | Sensor based truth maintenance | |
| Zheng et al. | Modeling taxi driver search behavior under uncertainty | |
| Lansley et al. | Challenges to representing the population from new forms of consumer data | |
| Yu et al. | Robust team orienteering problem with decreasing profits | |
| Li et al. | A probabilistic approach to detect mixed periodic patterns from moving object data | |
| US20140136280A1 (en) | Predictive Tool Utilizing Correlations With Unmeasured Factors Influencing Observed Marketing Activities | |
| CN110389963A (en) | The recognition methods of channel effect, device, equipment and storage medium based on big data | |
| Yap et al. | Aggregating multiple decision makers’ judgement | |
| Rui et al. | A location-dependent task assignment mechanism in vehicular crowdsensing | |
| Holtz | Limiting bias from test-control interference in online marketplace experiments | |
| Aunimo et al. | Big data governance in agile and data-driven software development: A market entry case in the educational game industry |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180330 |
|
| RJ01 | Rejection of invention patent application after publication |