CN108694625A - Equity preference predictor method, device and server - Google Patents
Equity preference predictor method, device and server Download PDFInfo
- Publication number
- CN108694625A CN108694625A CN201810708294.1A CN201810708294A CN108694625A CN 108694625 A CN108694625 A CN 108694625A CN 201810708294 A CN201810708294 A CN 201810708294A CN 108694625 A CN108694625 A CN 108694625A
- Authority
- CN
- China
- Prior art keywords
- equity
- user
- scene
- information
- estimated
- 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
- G06Q30/0201—Market modelling; Market analysis; Collecting market data
- G06Q30/0202—Market predictions or forecasting for commercial activities
Abstract
A kind of equity preference predictor method of this specification embodiment offer, device and server obtain environmental information and/or user behavior data that user is presently in environment in equity preference predictor method.According to environmental information and/or user behavior data, user's scene of user is determined.Obtain the scene information of user's scene.Algorithm is recalled according to predefined, recalls multiple equity to be estimated.The equity characteristic of multiple equity to be estimated is obtained, and obtains the user characteristic data of user.Scene information, equity characteristic and user characteristic data are inputted into equity preference prediction model, to estimate the preference score of multiple equity to be estimated.According to preference score, the target equity preferred to user is estimated.
Description
Technical field
This specification one or more embodiment is related to field of computer technology more particularly to a kind of equity preference side of estimating
Method, device and server.
Background technology
In order to increase client's number, promotion sales volume, expand the influence power of itself, businessman often carries out various marketing and lives
It is dynamic.Specifically, businessman can launch various interests to user and e.g. send telephone expenses, coupons, packet postal card and completely subtract voucher etc..
Traditional marketing activity is normally based on what scene on line was carried out.Such as, can be when user initiate demand, the request based on user
Condition estimates the equity preference of user.Rear line launch the equity of its preference.
However, most activities of user are at environment under line.Therefore, it is based on environment under line and carries out marketing activity
More and more paid attention to.Online under lower environment, user can't actively initiate demand.Therefore, how to being in ring under line
User's progress equity preference in border, which is estimated, just to be become and to solve the problems, such as.
Invention content
This specification one or more embodiment describes a kind of equity preference predictor method, device and server, can be with
Improve the accuracy that equity preference is estimated.
In a first aspect, a kind of equity preference predictor method is provided, including:
Obtain environmental information and/or user behavior data that user is presently in environment;
According to the environmental information and/or the user behavior data, user's scene of the user is determined;
Obtain the scene information of user's scene;
Algorithm is recalled according to predefined, recalls multiple equity to be estimated;
Obtain the equity characteristic of the multiple equity to be estimated and the user characteristic data of the user;
The scene information, the equity characteristic and the user characteristic data input equity preference are estimated into mould
Type, to estimate the preference score of the multiple equity to be estimated;
According to the preference score, the target equity preferred to the user is estimated.
Second aspect provides a kind of equity preference estimating device, including:
Acquiring unit is presently in the environmental information and/or user behavior data of environment for obtaining user;
Determination unit, the environmental information for being obtained according to the acquiring unit and/or the user behavior data,
Determine user's scene of the user;
The acquiring unit is additionally operable to obtain the scene information of user's scene;
Unit is recalled, for recalling algorithm according to predefined, recalls multiple equity to be estimated;
The acquiring unit is additionally operable to obtain the equity characteristic of the multiple equity to be estimated and the user
User characteristic data;
Input unit, the scene information, the equity characteristic for obtaining the acquiring unit and institute
User characteristic data input equity preference prediction model is stated, to estimate the preference score of the multiple equity to be estimated;
Unit is estimated, for according to the preference score, the target equity preferred to the user to be estimated.
The third aspect provides a kind of server, including:
Receiver is presently in the environmental information and/or user behavior data of environment for obtaining user;
At least one processor, for according to the environmental information and/or the user behavior data, determining the user
User's scene;Obtain the scene information of user's scene;Algorithm is recalled according to predefined, recalls and multiple waits estimating power
Benefit;The equity characteristic of the multiple equity to be estimated is obtained, and obtains the user characteristic data of the user;By the field
Scape information, the equity characteristic and the user characteristic data input equity preference prediction model, described more to estimate
The preference score of a equity to be estimated;According to the preference score, the target equity preferred to the user is estimated.
Equity preference predictor method, device and the server that this specification one or more embodiment provides obtain user
It is presently in the environmental information and/or user behavior data of environment.According to environmental information and/or user behavior data, determines and use
User's scene at family.Obtain the scene information of user's scene.Algorithm is recalled according to predefined, recalls multiple equity to be estimated.
The equity characteristic of multiple equity to be estimated is obtained, and obtains the user characteristic data of user.By scene information, equity feature
Data and user characteristic data input equity preference prediction model, to estimate the preference score of multiple equity to be estimated.According to
Preference score, the target equity preferred to user are estimated.Namely in this specification embodiment, carrying out, equity preference is pre-
When estimating, it is contemplated that the scene information for being presently in scene of user.Thus, it is possible to the accuracy that equity preference is estimated is improved, into
And precision of marketing under line can be improved.
Description of the drawings
It is required in being described below to embodiment to make in order to illustrate more clearly of the technical solution of this specification embodiment
Attached drawing is briefly described, it should be apparent that, the accompanying drawings in the following description is only some embodiments of this specification, right
For those of ordinary skill in the art, without creative efforts, it can also be obtained according to these attached drawings
Its attached drawing.
Fig. 1 is the equity preference Prediction System schematic diagram that this specification provides;
Fig. 2 is the equity preference predictor method flow chart that this specification one embodiment provides;
Fig. 3 is the equity preference estimating device schematic diagram that this specification one embodiment provides;
Fig. 4 is the server schematic diagram that this specification one embodiment provides.
Specific implementation mode
Below in conjunction with the accompanying drawings, the scheme provided this specification is described.
The equity preference predictor method that this specification one or more embodiment provides can be applied to power as shown in Figure 1
In beneficial preference Prediction System.In Fig. 1, equity preference Prediction System may include:Scene perception module 102, marketing channel decision
Module 104, collecting characterization data module 106 and equity recommending module 108.
Scene perception module 102 is used to be presently in the environmental information and/or user behavior data of environment according to user, really
Determine user's scene of user.Environmental information herein include but not limited to longitude and latitude location information, the network information (e.g., 2G, 3G,
WiFi titles and/address etc.) and facility information (e.g., operation system information and unit type etc.) etc..Above-mentioned user behavior
Data can refer to that user executes navigation patterns in intended application (application, APP), clicks behavior or consumption row
Generated data when to wait business conducts.Intended application herein can be the front-end software of equity preference Prediction System,
It can be used for interacting with user, including obtain information input by user and show information (e.g., notification message) to user.
Above-mentioned user's scene can pre-define, can include but is not limited to across city scene, into commercial circle scene, to shop
Recommend scene etc. on scape, payment associated scenario and line.
Specifically, it is (a kind of to incite somebody to action that the longitude and latitude location information that scene perception module 102 can be current to user carries out de-parsing
Longitude and latitude is parsed into an analytic method with semantic Text Address), to determine geographic area that user is presently in,
Such as, ×× country ×× saves ×× city.Obtain geographic area of the user residing for previous time point.If two geographic regions
Domain is inconsistent, it is determined that user's scene of user is across city scene.
Scene perception module 102 can also according to longitude and latitude location information and point of interest (Point of Interest,
POI) the boundary information in region, judges whether user enters the regions POI.The regions POI herein may include commercial circle, school or
Person hospital etc..The boundary information in the above-mentioned regions POI can be identified according to POI identification technologies from map datum.If
Into the regions POI, it is determined that user's scene of user is into commercial circle scene.
Scene perception module 102 can also be sentenced according to longitude and latitude location information, the network information and the location information in shop
Whether disconnected user enters shop.If into, it is determined that user's scene of user is to shop scene.It is wifi with the network information
For for title and/or address, the above-mentioned process for judging whether user enters shop can be:Skill can be identified by wifi
Art pre-establishes the correspondence in wifi titles and/or address and shop.It later, can be according to the wifi titles of active user
And/or address, corresponding shop is searched from above-mentioned correspondence.If found, judge that user enters shop.When
So, in practical applications, in order to improve accuracy, the longitude and latitude location information of user is can be combined with, whether to judge user
Enter shop.
Whether scene perception module 102 can also judge in user behavior data to include payment related information.If including
Then determine that user's scene of user is payment associated scenario.Payment related information herein can include but is not limited to payment gathering
The information of side and/or bill information etc..In addition, the operation of above-mentioned judgement is it can be appreciated that be the process that payment message is monitored.
Specifically, can include payment related information in corresponding user behavior data after user completes payment using payment platform.
It therefore, can be with according to whether comprising payment related information, to determine whether user's scene is payment associated scenario.It is being determined as propping up
After paying associated scenario, analysis mining can also be carried out to payment related information, to determine specific payment scene, e.g., in market
Scene, tour arrangement scene, viewing plan scene and the bill having a meal expire scene etc..
Whether scene perception module 102 can also judge in user behavior data to include equity relevant information.Power herein
Beneficial relevant information can include but is not limited to the term of validity etc. of the corresponding business of equity and equity.If including, it is determined that it uses
User's scene at family is to recommend scene on line.User can also log in APP and actively initiate recommendation request in Below-the-line.For example,
It logs in Alipay and checks that the Catering Pubs of recommendation are preferential, it is preferential etc. to check the arcade shop premises of recommendation.When user actively initiates to recommend
When request, user behavior data can include above-mentioned equity relevant information.
The scene perception module 102 that this specification provides, can pass through the environmental information and/or user behavior number of user
According to, the scene that user is presently in is accurately identified, it also can be independent of user's active request, so that this explanation
The scheme that book is provided can be adapted in the scene marketed under line.
It should be noted that scene perception module 102 is after determining user's scene of user, it can be by user's scene
Scene information and above-mentioned environmental information are sent to marketing channel decision-making module 104.Scene information herein may include but not
It is limited to user identifier (ID) and the relevant information etc. of a upper page.
Marketing channel decision-making module 104 is used to, according to predefined rule, target is chosen from least one dispensing channel
Launch channel.Dispensing channel herein can include but is not limited to short message, notice, the advertisement of bullet screen, top set advertisement and common wide
It accuses.Wherein, short message and notice are the dispensing channels that still can be touched in the case where user does not open intended application up to user, because
This, the rwo to bother degree most strong.
Above-mentioned predefined rule can be:After determining user's scene of user, guaranteeing to touch up to user's
It launches in channel, filters out the dispensing channel more than fatigue strength control limitation, degree dispensing channel as small as possible is bothered in selection.
For above-mentioned dispensing channel, in fatigue strength control aspect, the fatigue strength control for bothering the bigger dispensing channel of degree can be stringenter.
Specifically, it can be determined that whether intended application is opened.If it is not, then choosing fixed channel from least one dispensing channel.
Channel is launched using the fixation channel as target.If it is, determining each fatigue strength for launching channel and beating user
Disturb degree.According to scene information, fatigue strength and degree is bothered, target is chosen from least one dispensing channel and launches channel.
For example, when user is when Alipay is completed to pay, the advertisement position that pays successfully page is exactly best dispensing channel.In addition,
When user actively initiates recommendation request, the advertisement position of corresponding scene is exactly suitable dispensing channel.
The marketing channel decision-making module 104 that this specification provides, can select touch based on the scene residing for user
It reaches and bothers dispensing channel as small as possible so that the scheme that this specification is provided, can when applied to marketing scene under line
To select suitably to launch channel under current scene.
Target can be launched channel and above-mentioned by marketing channel decision-making module 104 after choosing target and launching channel
The scene information and environmental information of user's scene are sent to equity recommending module 108.
Before introducing equity recommending module 108, first collecting characterization data module 106 can be illustrated.Characteristic
According to collection module 106 for collecting user characteristic data and equity characteristic in advance.The collection process of user characteristic data can
Think:The essential information for collecting full dose user in advance, for example, age and gender etc..It later, can be to the essential information of user
And historical behavior information carries out feature mining, to obtain above-mentioned user characteristic data.The user characteristic data may include
The preference information etc. of user.Such as, user to the preferences of different type cuisines, to the preference etc. for the vehicles of going on a journey.Equity feature
The collection process of data can be:The essential information of different equity is collected, for example, the corresponding business of equity and equity is effective
Phase etc..Later, feature mining can be carried out to the essential information of equity, to obtain equity characteristic.The equity characteristic
According to the preferential dynamics and attraction etc. that can include but is not limited to equity.
Collecting characterization data module 106 can store the user characteristic data being collected into and equity characteristic to number
According in library.It should be noted that since the essential information of essential information/equity of user may change, it is possible to
Periodically to store in database user characteristic data and equity characteristic be updated.
Equity recommending module 108 is for estimating the preference equity of user.It may include walking as follows that this, which estimates process,
Suddenly:
1) algorithm is recalled according to predefined, recalls multiple equity to be estimated.It specifically, can be in the current location of user
Peripheral extent (e.g., N kilometer ranges) in recall multiple equity to be estimated.And/or it recalls respectively from multiple and different dimensions more
A equity to be estimated.Multiple dimension can include but is not limited to distance, using frequency and hot topic degree etc..And/or to
The historical behavior information at family is analyzed, to determine the interested equity of user.By being associated point to interested equity
It analyses to recall multiple equity to be estimated.In one implementation, above-mentioned interested equity can be expressed as vector come into
Row association analysis.After recalling multiple equity to be estimated, thick row's score of each equity to be estimated can be calculated.The thick row point
Several calculation formula is as follows:Thick row's score=apart from score * distance weightings+user behavior preference-score * preference weights+preferential power
Spend the preferential weights of score *.Wherein, it can refer to user apart from score and the distance between the shop value of equity to be estimated is provided.
User behavior score can be obtained after being analyzed the historical behavior information (e.g., consumer record or browsing record etc.) of user
It arrives.Preferential dynamics score can be determined according to the specific discount dynamics of equity to be estimated in history.Finally, according to thick row
Score is ranked up multiple equity to be estimated, and chooses forward M (e.g., the 200) equity to be estimated that sorts as candidate
Equity.
2) feature association.Feature association herein may include following two parts:First, from database obtain wait for it is pre-
Estimate equity or the candidate equity characteristic of equity and the user characteristic data of user.Second, scene information is carried out special
Sign association.Scene information can also include after feature association:Real time position, temporal information and weather etc..Wherein, real time position
Can carry out de-parsing to the longitude and latitude location information in environmental information to obtain.Such as, ×× city ×× region etc..Time
Information can parse to obtain to current time.What day it can include but is not limited to, to be what time, whether weekend and be
No is legal festivals and holidays etc..Weather can be arrived according to above-mentioned real time position and time inquiring.
3) equity preference is estimated.It can be treated based on advance trained equity preference prediction model and estimate equity or time
The preference score of equity is selected to be estimated.The equity preference prediction model can be the user characteristics according to user under real scene
Data, the equity characteristic of preferred equity and scene information, to logistic regression (Logistic Regression, LR)
Model and/or deep neural network (Deep Neural Network, DNN) model or gradient promote decision tree (Gradient
Boosting Decision Tree, GBDT) model obtains after being trained.It specifically, can be by the field after feature association
Scape information, user characteristic data and equity characteristic input equity preference prediction model, export equity or time to be estimated
Select the preference score of equity.
After the preference score for obtaining equity to be estimated or candidate equity, equity recommending module 108 can be according to inclined
Good grades are ranked up multiple equity to be estimated or candidate equity.For must have the scene of target equity return, power
Beneficial recommending module 108 can choose the n forward target equity that sort.The n target equity and target are launched into canal later
Road returns.For be not necessarily meant to recommendation results return scene, equity recommending module 108 may determine that equity to be estimated or
Whether the preference score of person's candidate's equity is higher than certain threshold value, chooses the m target equity that preference score is more than threshold value.Later will
The m target equity and target are launched channel and are returned.Certainly, in practical applications, it is also possible to choose less than more than threshold value
Target equity (i.e. m is 0).When choosing fall short equity, then terminate.
The equity recommending module 108 that this specification provides, can be by considering the feature association of scene information, and is based on power
Beneficial preference prediction model, the scene information of deep understanding user.So that the scheme that this specification provides is under applied to line
When scene of marketing, suitable equity under current scene can be selected to be presented to user.
Optionally, above-mentioned equity preference Prediction System can also include equity putting module 110.When further include equity launch
When module 110, the target equity and target of selection can be launched channel and return to equity dispensing mould by equity recommending module 108
Block 110.Later, channel is launched by target by equity putting module 110 and launches target equity.It is understood that executing
After above-mentioned dispensing operation, user terminal or target APP can show the target equity to user.
Fig. 2 is the equity preference predictor method flow chart that this specification one embodiment provides.The execution master of the method
Body can be the equipment with processing capacity:Either system or device can be e.g. that the equity preference in Fig. 1 is pre- to server
Estimate system.As shown in Fig. 2, the method can specifically include:
Step 202, environmental information and/or user behavior data that user is presently in environment are obtained.
Environmental information herein includes but not limited to longitude and latitude location information, the network information (e.g., 2G, 3G, WiFi title
With/address etc.) and facility information (e.g., operation system information and unit type etc.) etc..Above-mentioned user behavior data can be with
It refer to user's generated number when executing the business conducts such as navigation patterns, click behavior or consumer behavior in intended application
According to.
For above-mentioned environmental information, can be detect user in intended application and execute specified operation it is (e.g., downward
Slide) when obtain.Can be when user executes above-mentioned business conduct, by target for above-mentioned user behavior data
Using being recorded in local.Later, when the user behavior data of record is more than number of thresholds, equity preference Prediction System is uploaded to
Background data base in;Alternatively, can also be by being uploaded in background data base after the specified time of intended application interval.Therefore,
User behavior data can be obtained from background data base.
Step 204, according to environmental information and/or user behavior data, user's scene of user is determined.
Such as, can be that the use of user is determined according to environmental information and/or user behavior data by scene perception module 102
Family scene.User's scene herein can pre-define, and can include but is not limited to across city scene, into Shang Quanchang
Scape, to recommending scene etc. on shop scene, payment associated scenario and line.
Can be positioned to the longitude and latitude of the adjacent user collected twice by scene perception module 102 for across city scene
Information carries out de-parsing, and to determine two geographic areas of user, e.g., ×× country ×× saves ×× city.If this two
Geographic area is inconsistent, it is determined that user's scene of user is across city scene.
For entering commercial circle scene, the boundary letter that POI identification technologies identify the regions POI from map datum can be first passed through
Breath.The regions POI herein may include commercial circle, school or hospital etc..Later, according to longitude and latitude location information and the areas POI
The boundary information in domain, judges whether user enters the regions POI.If into the regions POI, it is determined that user's scene of user is
Into commercial circle scene.
For arriving shop scene, there is the wifi network of oneself in Most current shop.It can be by wifi identification technologies
Shop maps under wifi information (e.g., wifi titles and/or address) and line.Later, when the network information of acquisition includes
When wifi titles and/or address, it can be searched from above-mentioned mapping relations according to the wifi titles of active user and/or address
Corresponding shop.If found, judge that user enters shop.Certainly, in practical applications, in order to improve accuracy,
The longitude and latitude location information that can be combined with user, to judge whether user enters shop.
For paying associated scenario, payment message can be monitored.Specifically, when user completes payment using payment platform
Afterwards, can include payment related information in corresponding user behavior data.Payment related information herein may include but unlimited
In the information of payment beneficiary and/or bill information etc..It therefore, can be by whether judging in user behavior data comprising payment
Relevant information determines whether to listen to payment message.If listening to payment message, it is determined that user's scene is that payment is related
Scene.After being determined as paying associated scenario, analysis mining can also be carried out to payment related information, to determine specific payment
Scene e.g. expires scene etc. in scene, tour arrangement scene, viewing plan scene and the bill that market is had a meal.
For recommending scene on line, user can also log in intended application and actively initiate recommendation request in Below-the-line.Than
Such as, it logs in intended application and checks that the Catering Pubs of recommendation are preferential, it is preferential etc. to check the arcade shop premises of recommendation.When user actively initiates
When recommendation request, user behavior data can include equity relevant information.Equity relevant information herein may include but unlimited
In the corresponding business of equity and the term of validity etc. of equity.It therefore, can be by whether judging in user behavior data comprising power
Beneficial relevant information, to determine whether user's scene is to recommend scene on line.
Step 206, the scene information of user's scene is obtained.
Scene information herein can include but is not limited to user identifier (ID) and the relevant information etc. of a upper page.
Step 208, algorithm is recalled according to predefined, recalls multiple equity to be estimated.
Under normal conditions, the scale of equity is very big, and carrying out preference to ownership equity estimates, and can greatly influence systematicness
Energy.Therefore, equity can be first carried out before preference is estimated to recall, efficiently reduce equity scale.For the power marketed under line
Benefit, distance are very important factor, and then consider whether it is shop and whether be that user is preferred that user often goes
Preferential dynamics etc..The equity that this specification provides, which recalls step, to be:Peripheral extent (e.g., N in the current location of user
Kilometer range) in recall multiple equity to be estimated.And/or multiple equity to be estimated are recalled from multiple and different dimensions respectively.It should
Multiple dimensions can include but is not limited to distance, using frequency and hot topic degree etc..And/or the historical behavior of user is believed
Breath is analyzed, to determine the interested equity of user.Multiple wait for is recalled by being associated analysis to interested equity
Estimate equity.In one implementation, above-mentioned interested equity can be expressed as vector to be associated analysis.
Optionally, after recalling multiple equity to be estimated, thick row's score of each equity to be estimated can be calculated.This is thick
The calculation formula for arranging score is as follows:Thick row's score=apart from score * distance weightings+user behavior preference-score * preference weights+excellent
The preferential weights of favour dynamics score *.Wherein, it can refer to user apart from score and the distance between the shop of equity to be estimated is provided
Value.User behavior score can be analyzed the historical behavior information (e.g., consumer record or browsing record etc.) of user
It obtains afterwards.Preferential dynamics score can be determined according to the specific discount dynamics of equity to be estimated in history.Finally, according to
Thick row's score is ranked up multiple equity to be estimated, and chooses forward M (e.g., 200) equity conduct to be estimated of sorting
Candidate equity.
Step 210, the equity characteristic of multiple equity to be estimated and the user characteristic data of user are obtained.
The process of above-mentioned acquisition characteristic is it can be appreciated that be the process of feature association.When carrying out feature association,
Feature association can also be carried out to scene information.Detailed process can be:Longitude and latitude location information in environmental information is carried out
De-parsing obtains real time position.Such as, ×× city ×× region etc..Current time is parsed to obtain temporal information.Such as, week
It is several, be what time, whether weekend and whether be legal festivals and holidays etc..According to above-mentioned real time position and time inquiring to weather.By
Above it is found that scene information can also include after feature association:Real time position, temporal information and weather etc..
Can be received in advance by collecting characterization data module 106 for above-mentioned equity characteristic and user characteristic data
Collection is good and stores database.The collection process of user characteristic data can be:The essential information of full dose user is collected in advance,
For example, age and gender etc..Later, feature mining can be carried out to the essential information and historical behavior information of user, from
And obtain above-mentioned user characteristic data.The user characteristic data may include the preference information etc. of user.Such as, user is to inhomogeneity
The preference of type cuisines, to go on a journey the vehicles preference etc..The collection process of equity characteristic can be:Collect different equity
Essential information, for example, the term of validity etc. of the corresponding business of equity and equity.Later, can to the essential information of equity into
Row feature mining, to obtain equity characteristic.The equity characteristic can include but is not limited to the preferential dynamics of equity
And attraction etc..
It should be noted that features described above is to portray user with more rich information and wait estimating power the step of association
Benefit or candidate equity, so as to more accurately estimate the preference score of different equity.
Step 212, scene information, equity characteristic and user characteristic data are inputted into equity preference prediction model,
To estimate the preference score of multiple equity to be estimated.
When also carrying out feature association to scene information, the above-mentioned scene information for being input to equity preference prediction model can be with
It is the scene information after feature association.
Equity preference prediction model (also referred to as returning device) herein can be special according to the user of user under real scene
Data, the equity characteristic of preferred equity and scene information are levied, to logistic regression LR models and/or deep neural network
What DNN models or gradient promotion decision tree GBDT models obtained after being trained.
Step 214, according to preference score, the target equity preferred to user is estimated.
Specifically, sequence that can be according to preference score from high to low, to multiple equity to be estimated or candidate equity into
Row sequence.For must have the scene of target equity return, the n forward target equity that sort can be chosen.Later by the n
A target equity and target are launched channel (subsequently illustrating) and are returned.For being not necessarily meant to the field of recommendation results return
Scape, it can be determined that whether the preference score of equity to be estimated or candidate equity is higher than certain threshold value, chooses preference score and is more than
M target equity of threshold value.The m target equity and target channel is launched later to return.Certainly, in practical applications,
It may also choose less than the target equity (i.e. m is 0) more than threshold value.When choosing fall short equity, then terminate.
Channel is launched for above-mentioned target, can be by marketing channel decision-making module 104 according to predefined rule, to
It is chosen in a few dispensing channel.Dispensing channel herein can include but is not limited to short message, notice, the advertisement of bullet screen, top set
Advertisement and regular price-line advertising.Wherein, short message and notice still can be touched in the case where user does not open intended application up to use
The dispensing channel at family, therefore, the rwo to bother degree most strong.
Above-mentioned predefined rule can be:After determining user's scene of user, guaranteeing to touch up to user's
It launches in channel, filters out the dispensing channel more than fatigue strength control limitation, degree dispensing channel as small as possible is bothered in selection.
Under normal conditions, the fatigue strength control for bothering the bigger dispensing channel of degree can be stringenter.In one implementation, for tired
Lao Du is controlled, and can launch record and the feedback record of equity according to history to carry out.
Specifically, it can be determined that whether intended application is opened.If it is not, then being chosen from least one dispensing channel solid
Determine channel (e.g., short message and notice).Channel is launched using the fixation channel as target.If it is, determining each dispensing channel
Fatigue strength and degree is bothered to user.According to scene information, fatigue strength and degree is bothered, from least one dispensing canal
Target is chosen in road launches channel.For example, when user is when Alipay is completed to pay, the advertisement position for paying successfully page is exactly
Best dispensing channel.In addition, when user actively initiates recommendation request, the advertisement position of corresponding scene is exactly suitable dispensing canal
Road.
It should be noted that above-mentioned steps 208, step 210, step 212 and step 214 can recommend mould by equity
What block 108 executed.
To sum up, the equity preference predictor method that this specification embodiment provides, can perceive the scene that user is presently in,
Dispensing channel decisions, equity so as to triggering following are recommended and equity is launched, and equity is presented to user.In addition, this
The scheme that specification provides when perceiving user's scene, is carried out based on much information, e.g., real time position, the network of user
Information, user behavior data etc., so as to identify very more scenes.And it is also an option that rational dispensing channel.
Finally by deep understanding scene information, user's interested equity here and now is selected, so as to essence of marketing under increase line
Degree.
Accordingly with above-mentioned equity preference predictor method, a kind of equity preference that this specification one embodiment also provides is pre-
Device is estimated, as shown in figure 3, the device may include:
Acquiring unit 302 is presently in the environmental information and/or user behavior data of environment for obtaining user.
Determination unit 304, the environmental information for being obtained according to acquiring unit 302 and/or user behavior data, determine and use
User's scene at family.
User's scene herein may include one or more of:Across city scene, into commercial circle scene, to shop scene,
Recommend scene on payment associated scenario and line.
Acquiring unit 302 is additionally operable to obtain the scene information of user's scene.
Unit 306 is recalled, for recalling algorithm according to predefined, recalls multiple equity to be estimated.
Acquiring unit 302 is additionally operable to obtain the equity characteristic of multiple equity to be estimated and the user characteristics of user
Data.
Input unit 308, the scene information, equity characteristic for obtaining acquiring unit 302 and user characteristics
Data input equity preference prediction model, to estimate the preference score of multiple equity to be estimated.
Equity characteristic is by being obtained after treating the essential information progress feature mining for estimating equity.Equity feature
Data include the preferential dynamics and attraction of equity.User characteristic data is by the basic information and history row to user
For what is obtained after information progress feature mining.User characteristic data includes the preference information of user.
Equity preference prediction model is according to the user characteristic data of sample of users, scene information and preferred equity
Equity characteristic promotes decision tree GBDT moulds to logistic regression LR models and/or deep neural network DNN models or gradient
What type obtained after being trained.
Unit 310 is estimated, for according to preference score, the target equity preferred to user to be estimated.
Optionally, which can also include:
Selection unit 312, for according to predefined rule, choosing target dispensing canal from least one dispensing channel
Road.
Selection unit 312 specifically can be used for:
Judge whether intended application APP is opened.
If it is not, then choosing fixed channel from least one dispensing channel, fixed channel is launched into channel as target.
If it is, determining each fatigue strength for launching channel and bothering degree to user.According to scene information, tired
Labor degree and degree is bothered, target is chosen from least one dispensing channel and launches channel.
Unit 314 is launched, the target for being chosen by selection unit 312 launches channel and launches target equity.
Optionally, above-mentioned environmental information may include one or more of:Longitude and latitude location information and the network information.
Determination unit 304 specifically can be used for:
Pair warp and weft degree location information carries out de-parsing, to determine geographic area that user is presently in.User is obtained preceding
Geographic area residing for one time point, if two geographic areas are inconsistent, it is determined that user's scene of user is across city scene.
And/or
According to longitude and latitude location information and the boundary information in the regions POI, judge whether user enters the regions POI.Such as
Fruit enters, it is determined that user's scene of user is into commercial circle scene.
And/or
According to longitude and latitude location information and/or the network information and the location information in shop, judge whether user enters
Shop.If into, it is determined that user's scene of user is to shop scene.
And/or
Judge in user behavior data whether to include payment related information.If including, it is determined that user's scene of user
To pay associated scenario.
And/or
Judge in user behavior data whether to include equity relevant information.If including, it is determined that user's scene of user
To recommend scene on line.
Optionally, which can also include:
Computing unit 316, thick row's score for calculating multiple equity to be estimated.Thick row's score is to wait estimating according to multiple
Equity apart from score, user behavior preference-score and preferential dynamics score determine.
Selection unit 312, thick row's score for being calculated according to computing unit 316, chooses from multiple equity to be estimated
At least one candidate's equity.
Acquiring unit 302 specifically can be used for:
Obtain the equity characteristic of at least one candidate equity.
Optionally, unit 310 is estimated specifically to can be used for:
According to the sequence of preference score from high to low, at least one candidate equity is ranked up.
The forward candidate equity that will sort is chosen for target equity.Alternatively,
The candidate equity that preference score is more than threshold value is chosen from least one candidate equity.
The candidate equity for by preference score being more than threshold value is chosen for target equity.
Optionally, unit 306 is recalled specifically to can be used for:
Multiple equity to be estimated are recalled in the peripheral extent of the current location of user.And/or
Multiple equity to be estimated are recalled from multiple and different dimensions respectively, multiple dimensions include:Distance, using frequency with
And hot topic degree.And/or
The historical behavior information of user is analyzed, to determine the interested equity of user.By to interested power
Benefit is associated analysis to recall multiple equity to be estimated.The function of each function module of this specification above-described embodiment device,
Can be realized by each step of above method embodiment, therefore, this specification one embodiment provide device it is specific
The course of work does not repeat again herein.
The equity preference estimating device that this specification one embodiment provides, acquiring unit 302 obtain user and are presently in
The environmental information and/or user behavior data of environment.Determination unit 304 is determined according to environmental information and/or user behavior data
User's scene of user.Acquiring unit 302 obtains the scene information of user's scene.Unit 306 is recalled to be recalled according to predefined
Algorithm recalls multiple equity to be estimated.Acquiring unit 302 obtains equity characteristic and the user of multiple equity to be estimated
User characteristic data.Input unit 308 is pre- by scene information, equity characteristic and user characteristic data input equity preference
Model is estimated, to estimate the preference score of multiple equity to be estimated.Unit 310 is estimated according to preference score, it is preferred to user
Target equity is estimated.Thus, it is possible to improve the accuracy that equity preference is estimated, and then precision of marketing under line can be improved.
The equity preference estimating device that this specification one embodiment provides can be equity preference Prediction System in Fig. 1
One module or unit.
Accordingly with above-mentioned equity preference predictor method, this specification embodiment additionally provides a kind of server, such as Fig. 4 institutes
Show, which may include:
Receiver 402 is presently in the environmental information and/or user behavior data of environment for obtaining user.
At least one processor 404, for according to environmental information and/or user behavior data, determining the user of user
Scape.Obtain the scene information of user's scene.Algorithm is recalled according to predefined, recalls multiple equity to be estimated.Obtain multiple wait for
The equity characteristic of equity is estimated, and obtains the user characteristic data of user.By scene information, equity characteristic and use
Family characteristic inputs equity preference prediction model, to estimate the preference score of the multiple equity to be estimated.According to preference point
Number, the target equity preferred to user are estimated.
The server that this specification one embodiment provides, can improve the accuracy that equity preference is estimated, and then can be with
Improve precision of marketing under line.
Each embodiment in this specification is described in a progressive manner, identical similar portion between each embodiment
Point just to refer each other, and each embodiment focuses on the differences from other embodiments.Especially for server
For embodiment, since it is substantially similar to the method embodiment, so description is fairly simple, related place is implemented referring to method
The part explanation of example.
The step of method in conjunction with described in this disclosure content or algorithm can realize in a manner of hardware,
Can be that the mode of software instruction is executed by processor to realize.Software instruction can be made of corresponding software module, software
Module can be stored on RAM memory, flash memory, ROM memory, eprom memory, eeprom memory, register, hard
Disk, mobile hard disk, CD-ROM or any other form well known in the art storage medium in.A kind of illustrative storage Jie
Matter is coupled to processor, to enable a processor to from the read information, and information can be written to the storage medium.
Certainly, storage medium can also be the component part of processor.Pocessor and storage media can be located in ASIC.In addition, should
ASIC can be located in server.Certainly, pocessor and storage media can also be used as discrete assembly and be present in server.
Those skilled in the art are it will be appreciated that in said one or multiple examples, work(described in the invention
It can be realized with hardware, software, firmware or their arbitrary combination.It when implemented in software, can be by these functions
Storage in computer-readable medium or as on computer-readable medium one or more instructions or code be transmitted.
Computer-readable medium includes computer storage media and communication media, and wherein communication media includes convenient for from a place to another
Any medium of one place transmission computer program.It is any that storage medium can be that general or specialized computer can access
Usable medium.
It is above-mentioned that this specification specific embodiment is described.Other embodiments are in the scope of the appended claims
It is interior.In some cases, the action recorded in detail in the claims or step can be come according to different from the sequence in embodiment
It executes and desired result still may be implemented.In addition, the process described in the accompanying drawings not necessarily require show it is specific suitable
Sequence or consecutive order could realize desired result.In some embodiments, multitasking and parallel processing be also can
With or it may be advantageous.
Above-described specific implementation mode has carried out into one the purpose, technical solution and advantageous effect of this specification
Step is described in detail, it should be understood that the foregoing is merely the specific implementation mode of this specification, is not used to limit this
The protection domain of specification, all any modifications on the basis of the technical solution of this specification, made, change equivalent replacement
Into etc., it should all be included within the protection domain of this specification.
Claims (21)
1. a kind of equity preference predictor method, which is characterized in that including:
Obtain environmental information and/or user behavior data that user is presently in environment;
According to the environmental information and/or the user behavior data, user's scene of the user is determined;
Obtain the scene information of user's scene;
Algorithm is recalled according to predefined, recalls multiple equity to be estimated;
Obtain the equity characteristic of the multiple equity to be estimated and the user characteristic data of the user;
The scene information, the equity characteristic and the user characteristic data are inputted into equity preference prediction model,
To estimate the preference score of the multiple equity to be estimated;
According to the preference score, the target equity preferred to the user is estimated.
2. according to the method described in claim 1, it is characterized in that, further including:
According to predefined rule, target dispensing channel is chosen from least one dispensing channel;
Channel, which is launched, by the target launches the target equity.
3. according to the method described in claim 2, it is characterized in that, described according to predefined rule, from least one dispensing
Target is chosen in channel launches channel, including:
Judge whether intended application APP is opened;
If it is not, then choosing fixed channel from least one dispensing channel;Using the fixed channel as the target
Launch channel;
If it is, determining each fatigue strength for launching channel and bothering degree to user;According to the scene information, institute
State fatigue strength and it is described bother degree, from least one dispensing channel choose target launch channel.
4. according to claim 1-3 any one of them methods, which is characterized in that user's scene includes following a kind of or more
Kind:Across city scene, into commercial circle scene, to shop scene, payment associated scenario and line on recommend scene.
5. according to the method described in claim 4, it is characterized in that, the environmental information includes one or more of:Longitude and latitude
Spend location information and the network information;
It is described to determine user's scene of the user according to the environmental information and/or the user behavior data, including:
De-parsing, the geographic area being presently in the determination user are carried out to the longitude and latitude location information;Described in acquisition
Geographic area of the user residing for previous time point;If described two geographic areas are inconsistent, it is determined that the use of the user
Family scene is across the city scene;
And/or
According to the longitude and latitude location information and the boundary information in the regions POI, judge whether the user enters the POI
Region;If into, it is determined that user's scene of the user is described into commercial circle scene;
And/or
According to the longitude and latitude location information and/or the location information in the network information and shop, judge that the user is
It is no to enter the shop;If into, it is determined that user's scene of the user is described to shop scene;
And/or
Judge in the user behavior data whether to include payment related information;If including, it is determined that the user of the user
Scene is the payment associated scenario;
And/or
Judge in the user behavior data whether to include equity relevant information;If including, it is determined that the user of the user
Scene is to recommend scene on the line.
6. according to the method described in claim 1, it is characterized in that, the equity characteristic is by waiting estimating power to described
It is obtained after the essential information progress feature mining of benefit;The equity characteristic includes preferential dynamics and the attraction of equity
Power;The user characteristic data is carried out after feature mining by basic information to the user and historical behavior information
It arrives;The user characteristic data includes the preference information of user.
7. according to the method described in claim 1, it is characterized in that, special in the equity for obtaining the multiple equity to be estimated
Before levying data, further include:
Calculate thick row's score of the multiple equity to be estimated;Thick row's score be according to the multiple equity to be estimated away from
It is determined from score, user behavior preference-score and preferential dynamics score;
According to thick row's score, at least one candidate equity is chosen from the multiple equity to be estimated;
The equity characteristic for obtaining the multiple equity to be estimated includes:
Obtain the equity characteristic of at least one candidate equity.
8. the method according to the description of claim 7 is characterized in that described according to the preference score, partially to user institute
Good target equity is estimated, including:
According to the sequence of the preference score from high to low, at least one candidate equity is ranked up;
The forward candidate equity that will sort is chosen for the target equity;Alternatively,
The candidate equity that preference score is more than threshold value is chosen from least one candidate equity;
The candidate equity for by preference score being more than threshold value is chosen for the target equity.
9. according to the method described in claim 1, it is characterized in that, described recall algorithm according to predefined, multiple wait for is recalled
Equity is estimated, including:
Multiple equity to be estimated are recalled in the peripheral extent of the current location of the user;And/or
Respectively multiple equity to be estimated are recalled from multiple and different dimensions;Multiple dimensions include:Distance uses frequency and heat
Men Du;And/or
The historical behavior information of the user is analyzed, with the interested equity of the determination user;By to the sense
The equity of interest is associated analysis to recall multiple equity to be estimated.
10. according to the method described in claim 1, it is characterized in that, the equity preference prediction model is according to sample of users
User characteristic data, the equity characteristic of scene information and preferred equity, to logistic regression LR models and/or depth
What neural network DNN models or gradient promotion decision tree GBDT models obtained after being trained.
11. a kind of equity preference estimating device, which is characterized in that including:
Acquiring unit is presently in the environmental information and/or user behavior data of environment for obtaining user;
Determination unit, the environmental information for being obtained according to the acquiring unit and/or the user behavior data, determine
User's scene of the user;
The acquiring unit is additionally operable to obtain the scene information of user's scene;
Unit is recalled, for recalling algorithm according to predefined, recalls multiple equity to be estimated;
The acquiring unit is additionally operable to obtain the equity characteristic of the multiple equity to be estimated and the user of the user
Characteristic;
Input unit, the scene information, the equity characteristic for obtaining the acquiring unit and the use
Family characteristic inputs equity preference prediction model, to estimate the preference score of the multiple equity to be estimated;
Unit is estimated, for according to the preference score, the target equity preferred to the user to be estimated.
12. according to the devices described in claim 11, which is characterized in that further include:
Selection unit, for according to predefined rule, choosing target dispensing channel from least one dispensing channel;
Unit is launched, the target for being chosen by the selection unit launches channel and launches the target equity.
13. device according to claim 12, which is characterized in that the selection unit is specifically used for:
Judge whether intended application APP is opened;
If it is not, then choosing fixed channel from least one dispensing channel;Using the fixed channel as the target
Launch channel;
If it is, determining each fatigue strength for launching channel and bothering degree to user;According to the scene information, institute
State fatigue strength and it is described bother degree, from least one dispensing channel choose target launch channel.
14. according to claim 11-13 any one of them devices, which is characterized in that user's scene includes following one kind
Or it is a variety of:Across city scene, into commercial circle scene, to shop scene, payment associated scenario and line on recommend scene.
15. device according to claim 14, which is characterized in that the environmental information includes one or more of:Through
Latitude location information and the network information;
The determination unit is specifically used for:
De-parsing, the geographic area being presently in the determination user are carried out to the longitude and latitude location information;Described in acquisition
Geographic area of the user residing for previous time point;If described two geographic areas are inconsistent, it is determined that the use of the user
Family scene is across the city scene;
And/or
According to the longitude and latitude location information and the boundary information in the regions POI, judge whether the user enters the POI
Region;If into, it is determined that user's scene of the user is described into commercial circle scene;
And/or
According to the longitude and latitude location information and/or the location information in the network information and shop, judge that the user is
It is no to enter the shop;If into, it is determined that user's scene of the user is described to shop scene;
And/or
Judge in the user behavior data whether to include payment related information;If including, it is determined that the user of the user
Scene is the payment associated scenario;
And/or
Judge in the user behavior data whether to include equity relevant information;If including, it is determined that the user of the user
Scene is to recommend scene on the line.
16. according to the devices described in claim 11, which is characterized in that the equity characteristic is by waiting estimating to described
It is obtained after the essential information progress feature mining of equity;The equity characteristic includes preferential dynamics and the attraction of equity
Power;The user characteristic data is carried out after feature mining by basic information to the user and historical behavior information
It arrives;The user characteristic data includes the preference information of user.
17. according to the devices described in claim 11, which is characterized in that further include:
Computing unit, thick row's score for calculating the multiple equity to be estimated;Thick row's score is according to the multiple
Equity to be estimated apart from score, user behavior preference-score and preferential dynamics score determine;
Selection unit, thick row's score for being calculated according to the computing unit, is selected from the multiple equity to be estimated
Take at least one candidate equity;
The acquiring unit is specifically used for:
Obtain the equity characteristic of at least one candidate equity.
18. device according to claim 17, which is characterized in that the unit of estimating is specifically used for:
According to the sequence of the preference score from high to low, at least one candidate equity is ranked up;
The forward candidate equity that will sort is chosen for the target equity;Alternatively,
The candidate equity that preference score is more than threshold value is chosen from least one candidate equity;
The candidate equity for by preference score being more than threshold value is chosen for the target equity.
19. according to the devices described in claim 11, which is characterized in that the unit of recalling is specifically used for:
Multiple equity to be estimated are recalled in the peripheral extent of the current location of the user;And/or
Respectively multiple equity to be estimated are recalled from multiple and different dimensions;Multiple dimensions include:Distance uses frequency and heat
Men Du;And/or
The historical behavior information of the user is analyzed, with the interested equity of the determination user;By to the sense
The equity of interest is associated analysis to recall multiple equity to be estimated.
20. according to the devices described in claim 11, which is characterized in that the equity preference prediction model is according to sample of users
User characteristic data, the equity characteristic of scene information and preferred equity, to logistic regression LR models and/or depth
What neural network DNN models or gradient promotion decision tree GBDT models obtained after being trained.
21. a kind of server, which is characterized in that including:
Receiver is presently in the environmental information and/or user behavior data of environment for obtaining user;
At least one processor, for according to the environmental information and/or the user behavior data, determining the use of the user
Family scene;Obtain the scene information of user's scene;Algorithm is recalled according to predefined, recalls multiple equity to be estimated;It obtains
The equity characteristic of the multiple equity to be estimated is taken, and obtains the user characteristic data of the user;The scene is believed
Breath, the equity characteristic and the user characteristic data input equity preference prediction model, to estimate the multiple wait for
Estimate the preference score of equity;According to the preference score, the target equity preferred to the user is estimated.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810708294.1A CN108694625A (en) | 2018-07-02 | 2018-07-02 | Equity preference predictor method, device and server |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810708294.1A CN108694625A (en) | 2018-07-02 | 2018-07-02 | Equity preference predictor method, device and server |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108694625A true CN108694625A (en) | 2018-10-23 |
Family
ID=63850286
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810708294.1A Pending CN108694625A (en) | 2018-07-02 | 2018-07-02 | Equity preference predictor method, device and server |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108694625A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110046736A (en) * | 2018-12-20 | 2019-07-23 | 阿里巴巴集团控股有限公司 | Equity recommended method, device and equipment |
| CN111415178A (en) * | 2019-01-04 | 2020-07-14 | 阿里巴巴集团控股有限公司 | User rights information providing method and device and electronic equipment |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102158802A (en) * | 2011-02-15 | 2011-08-17 | 广州市动景计算机科技有限公司 | Information distribution method and device |
| TW201333846A (en) * | 2011-09-30 | 2013-08-16 | Rakuten Inc | Payment system and payment terminal |
| CN103282926A (en) * | 2010-12-23 | 2013-09-04 | Cje&M公司 | System and method for notifying and providing a coupon using sns information |
| CN104679743A (en) * | 2013-11-26 | 2015-06-03 | 阿里巴巴集团控股有限公司 | Method and device for determining preference model of user |
| CN106021507A (en) * | 2016-05-20 | 2016-10-12 | 广东欧珀移动通信有限公司 | Application recommendation method and apparatus, and mobile terminal |
| CN106209959A (en) * | 2015-05-26 | 2016-12-07 | 徐尚英 | Network service intelligence based on user's request finds method |
| CN106599137A (en) * | 2016-12-02 | 2017-04-26 | 北京薇途科技有限公司 | Novel scene content pushing system and device |
| CN107730038A (en) * | 2017-10-09 | 2018-02-23 | 小草数语（北京）科技有限公司 | The other Forecasting Methodology of user preference, device and its equipment |
| CN107766462A (en) * | 2017-09-28 | 2018-03-06 | 重庆大学 | Point of interest based on user preference, social credit worthiness and geographical position recommends method |
| CN107818478A (en) * | 2017-10-31 | 2018-03-20 | 携程计算机技术（上海）有限公司 | Overseas reward voucher distribution method and system based on provisional profit |
| CN107895213A (en) * | 2017-12-05 | 2018-04-10 | 北京三快在线科技有限公司 | Forecasting Methodology, device and the electronic equipment of spending limit |
-
2018
- 2018-07-02 CN CN201810708294.1A patent/CN108694625A/en active Pending
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103282926A (en) * | 2010-12-23 | 2013-09-04 | Cje&M公司 | System and method for notifying and providing a coupon using sns information |
| CN102158802A (en) * | 2011-02-15 | 2011-08-17 | 广州市动景计算机科技有限公司 | Information distribution method and device |
| TW201333846A (en) * | 2011-09-30 | 2013-08-16 | Rakuten Inc | Payment system and payment terminal |
| CN104679743A (en) * | 2013-11-26 | 2015-06-03 | 阿里巴巴集团控股有限公司 | Method and device for determining preference model of user |
| CN106209959A (en) * | 2015-05-26 | 2016-12-07 | 徐尚英 | Network service intelligence based on user's request finds method |
| CN106021507A (en) * | 2016-05-20 | 2016-10-12 | 广东欧珀移动通信有限公司 | Application recommendation method and apparatus, and mobile terminal |
| CN106599137A (en) * | 2016-12-02 | 2017-04-26 | 北京薇途科技有限公司 | Novel scene content pushing system and device |
| CN107766462A (en) * | 2017-09-28 | 2018-03-06 | 重庆大学 | Point of interest based on user preference, social credit worthiness and geographical position recommends method |
| CN107730038A (en) * | 2017-10-09 | 2018-02-23 | 小草数语（北京）科技有限公司 | The other Forecasting Methodology of user preference, device and its equipment |
| CN107818478A (en) * | 2017-10-31 | 2018-03-20 | 携程计算机技术（上海）有限公司 | Overseas reward voucher distribution method and system based on provisional profit |
| CN107895213A (en) * | 2017-12-05 | 2018-04-10 | 北京三快在线科技有限公司 | Forecasting Methodology, device and the electronic equipment of spending limit |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110046736A (en) * | 2018-12-20 | 2019-07-23 | 阿里巴巴集团控股有限公司 | Equity recommended method, device and equipment |
| CN110046736B (en) * | 2018-12-20 | 2023-12-12 | 创新先进技术有限公司 | Rights and interests recommending method, device and equipment |
| CN111415178A (en) * | 2019-01-04 | 2020-07-14 | 阿里巴巴集团控股有限公司 | User rights information providing method and device and electronic equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Hu et al. | Your neighbors affect your ratings: on geographical neighborhood influence to rating prediction | |
| CN106991576B (en) | Method and device for displaying heat of geographic area | |
| US10552504B1 (en) | Selecting, ranking, and/or presenting microsite content | |
| CN104731917B (en) | A kind of recommendation method and device | |
| Wang et al. | Understanding travellers’ preferences for different types of trip destination based on mobile internet usage data | |
| US9183497B2 (en) | Performance-efficient system for predicting user activities based on time-related features | |
| CN102591911B (en) | The real time individual of position related entities is recommended | |
| US8417683B2 (en) | System and method for presenting geolocated relevance-based content | |
| CN110147967B (en) | Risk prevention and control method and device | |
| CN106911801B (en) | method for associating user information and information pushing system | |
| US20160012512A1 (en) | Lifestyle recommendation system | |
| KR101062927B1 (en) | Method, system and computer-readable recording medium for recommending other users or objects by considering at least one user's preference | |
| CN108596695B (en) | Entity pushing method and system | |
| US20160063597A1 (en) | Product Recommendation Based On Geographic Location And User Activities | |
| US8631158B1 (en) | Location-based automatic modification of a portable browser's home page and bookmarks | |
| CN112241489B (en) | Information pushing method, device, readable storage medium and computer equipment | |
| US20210224851A1 (en) | Affiliate-driven benefits matching system and methods with location-triggered benefit alert and search score determination | |
| CN111080339A (en) | Method and device for generating category preference data based on scene | |
| CN108694625A (en) | Equity preference predictor method, device and server | |
| US10545028B1 (en) | System and method of generating route-based ad networks | |
| US20170016736A1 (en) | Notification of proximal points of interest | |
| WO2012109175A2 (en) | Opportunity identification for search engine optimization | |
| KR101732873B1 (en) | Method and system for recommending user custmized popular place | |
| CN110659920B (en) | Method and device for recommending service object | |
| CN108460630B (en) | Method and device for carrying out classification analysis based on user data |
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 | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20200930 Address after: Cayman Enterprise Centre, 27 Hospital Road, George Town, Grand Cayman Islands Applicant after: Advanced innovation technology Co.,Ltd. Address before: A four-storey 847 mailbox in Grand Cayman Capital Building, British Cayman Islands Applicant before: Alibaba Group Holding Ltd. Effective date of registration: 20200930 Address after: Cayman Enterprise Centre, 27 Hospital Road, George Town, Grand Cayman Islands Applicant after: Innovative advanced technology Co.,Ltd. Address before: Cayman Enterprise Centre, 27 Hospital Road, George Town, Grand Cayman Islands Applicant before: Advanced innovation technology Co.,Ltd. |
|
| TA01 | Transfer of patent application right | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20181023 |
|
| RJ01 | Rejection of invention patent application after publication |