CN110046928A - Determine method, apparatus, electronic equipment and the readable storage medium storing program for executing of label - Google Patents
Determine method, apparatus, electronic equipment and the readable storage medium storing program for executing of label Download PDFInfo
- Publication number
- CN110046928A CN110046928A CN201910168755.5A CN201910168755A CN110046928A CN 110046928 A CN110046928 A CN 110046928A CN 201910168755 A CN201910168755 A CN 201910168755A CN 110046928 A CN110046928 A CN 110046928A
- Authority
- CN
- China
- Prior art keywords
- user
- behavior
- probability
- data
- label
- 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
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/067—Enterprise or organisation modelling
-
- 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
Landscapes
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Strategic Management (AREA)
- Entrepreneurship & Innovation (AREA)
- Development Economics (AREA)
- Human Resources & Organizations (AREA)
- Accounting & Taxation (AREA)
- Finance (AREA)
- Economics (AREA)
- Game Theory and Decision Science (AREA)
- Marketing (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Educational Administration (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The embodiment of the present application provides method, apparatus, electronic equipment and the readable storage medium storing program for executing of a kind of determining label, to improve the accuracy for determining label.The described method includes: determining that the behavior of estimating of the user generates probability according to the historical behavior data of user;The behavior of estimating of the user is generated into probability input calibrating patterns, it is calibrated with the behavior generation probability of estimating to the user, wherein, the calibrating patterns are to estimate behavior with multiple groups to generate the mapping relations between probability sample value and agenda generation probability sample value to input, and are trained obtained model to the first preset model;Probability is generated according to the behavior of estimating after the calibration of the user, determines the label of the user.
Description
Technical field
The invention relates to technical field of data processing more particularly to a kind of method, apparatus of determining label, electronics
Equipment and readable storage medium storing program for executing.
Background technique
The label for accurately determining user is of great significance to information push, can be accurately according to the label of user
Suitable information is pushed to user.
The relevant technologies provide a kind of method for determining user tag based on user's portrait.This method is used, firstly, determining
The user of user draws a portrait, and determines user's portrait of a user and the information that uses includes the essential information of the user, such as: property
Not, age, constellation, permanent residence etc. also include the behavior frequency of the user, such as: the Information frequency, information browse frequency etc.
Deng.Then, it is drawn a portrait according to the user of a user and determines the label of the user.
However, the method for the label of above-mentioned determining user depends only on user's portrait, and determine user's portrait of user
Information it is not necessarily true, the accuracy for causing user to draw a portrait is not high enough, and then directly influences the accuracy of user tag.Cause
And the accuracy of the method for determination user tag in the related technology is to be improved.
Summary of the invention
The embodiment of the present application provides method, apparatus, electronic equipment and the readable storage medium storing program for executing of a kind of determining label, to improve
Determine the accuracy of the label of user.
The embodiment of the present application first aspect provides a kind of method of the label of determining user, which comprises
According to the historical behavior data of user, determine that the behavior of estimating of the user generates probability;
The behavior of estimating of the user is generated into probability input calibrating patterns, is generated generally with the behavior of estimating to the user
Rate is calibrated, wherein the calibrating patterns are to estimate behavior with multiple groups to generate probability sample value and agenda generation probability
Mapping relations between sample value are input, are trained obtained model to the first preset model;
Probability is generated according to the behavior of estimating after the calibration, determines the label of the user.Optionally, the method is also
Include:
Obtain the target component of information publisher setting;
Acquire the user with the associated historical behavior data of the target component.
Optionally, after the label for determining the user, the method also includes:
According to the target component, target audience is determined;
The information that the information publisher issues is pushed to the target audience.
Optionally, the calibrating patterns obtain according to the following steps:
According to the multiple sample data it is respective estimate behavior generate probability size, to the multiple sample data into
Row sequence;
Multiple sample data graduation after sequence are divided into multiple groups sample data;
For every group of sample data in the multiple groups sample data, probability is generated to the behavior of estimating of this group of sample data
It is averaged, obtains the group and estimate behavior generation probability sample value;
Mark whether it is preset mark according to what sample data each in this group of sample data carried, determines this group of sample number
Label according to middle carrying is the accounting of the sample data of preset mark, and the accounting is determined as this group of agenda and is generated generally
Rate sample value;
The mapping relations between behavior generation probability sample value and agenda generation probability sample value are estimated according to multiple groups
For input, the first preset model is trained, the calibrating patterns are obtained.
Optionally, according to the historical behavior data of user, determine that the behavior of estimating of the user generates probability, comprising:
Obtain historical behavior data and outer historical behavior data of standing in the station of the user;
It is to generate probability to estimate mould by historical behavior data in the station of the user and outer historical behavior data input columns of standing
Type determines that the behavior of estimating of the user generates probability.
Optionally, according to the historical behavior data of user, determine that the behavior of estimating of the user generates probability, comprising:
It is to generate probability prediction model by the historical behavior data input columns of the user, with estimating for the determination user
Behavior generates probability, wherein it is with multiple historical behavior sample datas for carrying label that the behavior, which generates probability prediction model,
For input, obtained model is trained to the second preset model, wherein the label that each historical behavior data carry
Whether characterization user makes agenda.
The embodiment of the present application second aspect provides a kind of device of determining label, and described device includes:
First determining module determines that the behavior of estimating of the user generates generally for the historical behavior data according to user
Rate;
Calibration module, for the behavior of estimating of the user to be generated probability input calibrating patterns, to the user's
It estimates behavior generation probability to be calibrated, wherein the calibrating patterns are to estimate behavior with multiple groups to generate probability sample value and reality
It is input that border behavior, which generates the mapping relations between probability sample value, is trained obtained model to the first preset model;
Second determining module determines label for generating probability according to the behavior of estimating after the calibration for the user.
Optionally, described device further include:
First obtains module, for obtaining the target component of information publisher setting;
Acquisition module, for acquiring the user and the associated historical behavior data of the target component.
Optionally, described device further include:
Third determining module, for after the label for determining the user, according to the target component, determine target by
It is many;
Pushing module, for after the label for determining the user, Xiang Suoshu target audience to push the information publication
The information of Fang Fabu.
Optionally, described device further include:
Sorting module, for according to the multiple sample data it is respective estimate behavior generate probability size, to described
Multiple sample datas are ranked up;
Division module, for multiple sample data graduation after sequence to be divided into multiple groups sample data;
Computing module, for for every group of sample data in the multiple groups sample data, to the pre- of this group of sample data
Estimate behavior generation probability to be averaged, obtains the group and estimate behavior generation probability sample value;
4th determining module marks whether it is pre- bidding for what is carried according to sample data each in this group of sample data
Note, determines that the label carried in this group of sample data is the accounting of the sample data of preset mark, and the accounting is determined as
This group of agenda generates probability sample value.
Training module, for according to multiple groups estimate behavior generate probability sample value and agenda generate probability sample value it
Between mapping relations be input, the first preset model is trained, the calibrating patterns are obtained.
Optionally, first determining module includes:
Second obtains module, historical behavior data and outer historical behavior data of standing in the station for obtaining the user;
Input module, for being to produce by historical behavior data in the station of the user and outer historical behavior data input columns of standing
Raw probability prediction model determines that the behavior of estimating of the user generates probability.
Optionally, first determining module includes:
Input submodule, for being to generate probability prediction model by the historical behavior data input columns of the user, with true
The behavior of estimating of the fixed user generates probability, wherein it is with multiple carryings label that the behavior, which generates probability prediction model,
Historical behavior sample data is input, is trained obtained model to the second preset model, wherein each history row
Characterize whether user makes agenda for the label that data carry.
The embodiment of the present application third aspect provides a kind of computer readable storage medium, is stored thereon with computer program,
The step in the method for calibrating label really as described in the application first aspect is realized when the program is executed by processor.
The embodiment of the present application fourth aspect provides a kind of electronic equipment, including memory, processor and is stored in memory
Computer program that is upper and can running on a processor, the processor realize determination described in the application first aspect when executing
The step of method of label.
Using a kind of method of determining label provided by the embodiments of the present application, pre-establishes and estimate row for calibrate user
For generate probability calibrating patterns, obtain multiple users estimate behavior generate probability after, be not directly according to multiple users
Behavior of estimating generate probability to determine the label of user, but the calibrating patterns pre-established are utilized, to the pre- of multiple users
Estimate behavior generation probability to be calibrated, probability is then generated according to the behavior of estimating after the calibration of multiple users, determines user's
Label.Due to determining that the label of user is carried out according to the behavior generation probability of estimating after calibrating, that is, use is more accurate
The label that behavior generates determine the probability user is estimated, so improving the accuracy of the label of determining user.
Detailed description of the invention
Technical solution in ord to more clearly illustrate embodiments of the present application, below by institute in the description to the embodiment of the present application
Attached drawing to be used is needed to be briefly described, it should be apparent that, the accompanying drawings in the following description is only some implementations of the application
Example, for those of ordinary skill in the art, without any creative labor, can also be according to these attached drawings
Obtain other attached drawings.
Fig. 1 is that one embodiment of the application proposes the flow chart for calibrating the method for label really;
Fig. 2 is the flow chart of the method for the acquisition historical behavior data that one embodiment of the application proposes;
Fig. 3 is the flow chart for the method to user's recommendation information that one embodiment of the application proposes;
Fig. 4 is the flow chart of the determination user that one embodiment of the application proposes estimated behavior and generate probability;
Fig. 5 is the flow chart of the method for the training calibrating patterns that one embodiment of the application proposes;
Fig. 6 is another flow chart of the method for the acquisition historical behavior data that another embodiment of the application provides;
Fig. 7 is the flow chart for the method that the advertisement that another embodiment of the application provides is launched;
Fig. 8 is that one embodiment of the application proposes the schematic diagram for calibrating the device of label really;
Fig. 9 is the schematic diagram for the electronic equipment that one embodiment of the application proposes.
Specific embodiment
Below in conjunction with the attached drawing in the embodiment of the present application, technical solutions in the embodiments of the present application carries out clear, complete
Site preparation description, it is clear that described embodiment is some embodiments of the present application, instead of all the embodiments.Based on this Shen
Please in embodiment, every other implementation obtained by those of ordinary skill in the art without making creative efforts
Example, shall fall in the protection scope of this application.
It is that one embodiment of the application proposes the flow chart for calibrating the method for label really with reference to Fig. 1, Fig. 1.As shown in Figure 1, should
Method the following steps are included:
Step S11: according to the historical behavior data of user, determine that the behavior of estimating of the user generates probability.
In the present embodiment, the historical behavior data of user include but is not limited to: in the station of user historical behavior data and
It stands outer historical behavior data.Wherein, " in standing " and " stand outer " is for particular platform (such as: electric business platform).With
The historical behavior data at family include but is not limited to following two categories information: type I information is historical behavior number in the station of the user
According to, such as: the browsing behavior of the user, click behavior, logs in behavior etc. at collection behavior;Second category information is the user's
It stands outer historical behavior data, such as: the model for the electronic equipment that the title for the client that the user installation is crossed, the user use,
The advertisement etc. that the user clicked.Optionally, the historical behavior data of user can also in conjunction with the essential information of user, with
Probability is generated convenient for the behavior of estimating of subsequent determining user.Wherein, the essential information of user include: gender, the age, constellation, often
Guard station etc..
Wherein, it estimates behavior and generates what the information that " behavior " in probability refers to that user recommends for particular platform was made
Behavior, such as when the information of recommendation is an advertising information, user can make click behavior to the advertising information, collection is gone
For, splitting glass opaque etc..It estimates behavior generation probability to refer to before the information recommendation that will recommend is to user, pre-estimates user couple
The information of the recommendation generates the probability of each behavior, such as: pre-estimate the probability that user clicks an advertising information, Huo Zheshou
Hide the probability of an advertising information.
According to the historical behavior data of user, it can determine that the behavior generation probability of estimating of the user refers to: by user
Historical behavior data input columns be generate probability prediction model (behavior generate probability prediction model concept hereinafter say
It is bright).
The a plurality of sample data of multiple users can be obtained according to the historical behavior data of user.In each implementation of the application
In example, all information for recommending user can form the sample data for some user, and user is to all samples
Notebook data not necessarily produces user behavior, such as: user views the information of (or not viewing) recommendation still
The user behaviors such as click, collection or sharing are not made, and still, the information that user recommends for this can also form one
Sample data.
The sample data of user can be about some particular user behavior, such as: the sample number about the behavior of click
According to sample data, or the sample data about splitting glass opaque etc. about collection behavior.Sample data about the behavior of click
Refer to: selecting the sample data of a part of user, if this certain customers is calculated respectively according to this part sample data
Clicking rate, then using this part sample data as about click behavior sample data.Similarly, select a part of user's
Sample data, if the respective collection rate of this certain customers to be calculated according to this part sample data, by this part
Sample data is as the sample data about collection behavior.
The sample data of one user may include: the ID of the user, according to the feedback of user (for the letter of recommendation
Breath, user is made that user behavior or do not make user behavior) (additional information can be from for the data that generate and additional information
It is arbitrarily determined in the historical behavior data and essential information of the user, such as: gender, age, occupation, constellation etc., it can also basis
The requirement of information publisher determines).Such as: a sample data about the behavior of click specifically may is that a user clicks
The information of one recommendation and the sample data or a user generated is not clicked the information of a recommendation and is given birth to
At a sample data.
For the information that same is recommended, multiple and different users can form a plurality of different sample data, will be each
Sample data input behavior generates probability prediction model, can predict each user corresponding with sample data and push away for this
The information recommended makes the probability of user behavior.For all information that particular platform is recommended, the same user can form a plurality of
Each sample data input behavior is generated probability prediction model by sample data, can predict the user for each
The information of recommendation makes the probability of user behavior.
Step S12: the behavior of estimating of the user is generated into probability input calibrating patterns, to estimate row to the user
It is calibrated to generate probability, wherein the calibrating patterns are to estimate behavior with multiple groups to generate probability sample value and agenda
The mapping relations generated between probability sample value are input, are trained obtained model to the first preset model.
In the present embodiment, in order to ensure the behavior of estimating for generating the user that probability prediction model obtains by behavior generates
Probability more approaching to reality probability is calibrated using behavior generation probability of estimating of the calibrating patterns to user.
It is that the user predicted a user pushes away for one that one group, which is estimated behavior to generate probability sample value,
The information recommended makes the probability of user behavior, such as: one group estimate behavior generate probability sample value about click behavior when, the sample
This value is higher, and the clicking rate for representing the user that prediction obtains is higher, and the sample value is lower, represents the user's that prediction obtains
Clicking rate is lower.Estimating behavior with one group, to generate probability sample value corresponding, is one group of agenda generation probability sample value,
Indicate that the user is directed to the practical probability for making user behavior of information of a recommendation.One user is corresponding with one group and estimates behavior
It generates probability sample value and one group of agenda generates probability sample value, multiple users are corresponding with multiple groups and estimate behavior generation probability
Sample value and multiple groups agenda generate probability sample value, estimate behavior with multiple groups and generate probability sample value and multiple groups agenda
The mapping relations between probability sample value are generated as input, the first preset model are trained to get calibrating patterns are arrived.
Probability sample value is generated according to behavior of estimating and agenda generates the training of probability sample value and obtains calibrating patterns
Detailed process will be described hereinafter.
Step S13: probability is generated according to the behavior of estimating after the calibration, determines the label of the user.
In the present embodiment, after the calibration obtained for a user in predicting estimate behavior generate probability it is higher, represent
The probability that the user makes user behavior to the information of recommendation is higher.Behavior of estimating after the calibration of multiple users is generated into probability
It is divided into multiple groups in a certain order, and to the label of each group of addition user, pipe easily can be carried out to user
Reason.The case where user makes user behavior for the information recommended can be viewed according to the label of user.
By taking user behavior is click behavior as an example, the behavior of estimating about the behavior of click after multiple calibrations is generated into probability
Multiple groups are divided into according to certain rule, such as is arranged according to descending, and user of the probability value between 80%-100% is drawn
It is divided into A group, user of the probability value between 60%-80% is divided into B group, by user of the probability value between 40%-60%
It is divided into C group, and so on, obtain multiple groups.It then is the labels for dividing multiple groups obtained and successively adding user, such as:
Label for the user of A group addition is " the very high user group of probability for clicking the information recommended ", for the user of B group addition
Label be " click recommend information the higher user group of probability ", be C group addition user label be " click recommendation
Information the lower user group of probability ".
The same user can produce multiple and different user behaviors, and according to the difference of user behavior, the same user can
To belong to multiple and different labels.Such as user behavior is that be formed when click behavior is about click behavior in above-described embodiment
User label, when user behavior is other types, for example when collection behavior, which may belong to one about collection again
The label of behavior, or, when user behavior is splitting glass opaque, which may belong to a mark about splitting glass opaque again
Label.
Under different labels, the probability that user makes the corresponding user behavior of the label be can be different.Such as: it is same
One user, the probability for making click behavior to the information of recommendation can make the general of collection behavior between 80%-100%
Rate can be between 60%-80%, and the probability for making splitting glass opaque can be between 40%-60% etc..
Above-mentioned step S11 to step S13 is described in detail with a specific embodiment below.Finally to obtain
Label be about click behavior label for, detailed process are as follows:
The first step, obtain multiple groups about click behavior sample data (one group about click behavior sample data be root
According to a plurality of sample data that the information of some or all recommendations of a user generates, such as: for recommending the complete of the user
Portion's history recommendation information, can obtain one group belong to the user about click behavior sample data), then every group about
In the sample data of click behavior, every sample data input behavior about the behavior of click is generated into probability prediction model, really
This fixed estimates behavior about one of the corresponding user of sample data of the behavior of click and generates the probability (letter that each is recommended
Breath usually can all measure the probability that each user clicks the information of the recommendation in advance, that is, estimate behavior before recommending user
Generate probability).
Second step, the behavior of estimating by multiple groups about the behavior of click generates probability input calibrating patterns, to wherein every group
Behavior of estimating about the behavior of click generates probability and is calibrated, and estimates row to obtain the approaching to reality probability value of each user
To generate probability.
The behavior of estimating of third step, the approaching to reality probability value of multiple users according to obtained in second step generates probability,
Group's division is carried out to multiple users, and label is set for each group, wherein according to each label, can obtain and belong to this
The probability for the information that the click of the user of label is recommended.
If the label finally obtained is the label about collection behavior, first have to obtain multiple groups about collection behavior
Sample data;If the label finally obtained is the label about splitting glass opaque, first have to obtain multiple groups about splitting glass opaque
Sample data, and so on.
Using a kind of method of determining label provided by the embodiments of the present application, pre-establishes and estimate row for calibrate user
For generate probability calibrating patterns, obtain multiple users estimate behavior generate probability after, be not directly according to multiple users
Behavior of estimating generate probability to determine the label of user, but the calibrating patterns pre-established are utilized, to the pre- of multiple users
Estimate behavior generation probability to be calibrated, probability is then generated according to the behavior of estimating after the calibration of multiple users, determines user's
Label.Due to determining that the label of user is carried out according to the behavior generation probability of estimating after calibrating, that is, use is more accurate
The label that behavior generates determine the probability user is estimated, so improving the accuracy of the label of determining user.
It is the flow chart of the method for the acquisition historical behavior data that one embodiment of the application proposes with reference to Fig. 2, Fig. 2.It is holding
Before row step S11- step S13, it is also necessary to obtain historical behavior data.As shown in Fig. 2, method includes the following steps:
Step S21: the target component of information publisher setting is obtained;
Step S22: acquire the user with the associated historical behavior data of the target component.
First in the step s 21, it is the item for reaching certain information recommendation effect and being arranged that target component, which is information publisher,
Part, target component can be the condition for user behavior, such as: the clicking rate of desired user, sharing rate, turns collection rate
Rate etc.;Or be for user essential information restrictive condition, such as: the gender of desired target user, age
Section, occupation etc.;It can also be other conditions, such as: the specific period.The purpose of target component is set, is to improve
The effect of final information recommendation.
For step S22, since the historical behavior data bulk of user is more, therefore, it is not necessary to must disposably be gone through to all
History behavioral data is trained, and can pointedly usage history behavioral data.How pointedly to use, it is necessary to according to
The target component of user setting is screened in all historical behavior data.
Such as when target component is the interval range of a clicking rate, historical behavior number associated with the target component
According to including at least: normal condition (such as: there is not user log off account, that particular platform is not used for a long time situations such as)
Under a part of history recommendation information for recommending user;It is related to the target component when target component is the gender of user
The historical behavior data of connection include at least: the recommendation information of a part of history for recommending female user under normal circumstances.
In the present embodiment, not directly using all historical behavior data of user as sample data, but pass through
The screening of target component, then using the historical behavior data after screening as the sample data of user, can make later use should
The model that sample data training after screening obtains more has specific aim, more can neatly cope with information publisher proposition
Various requirement.
It is the flow chart for the method to user's recommendation information that one embodiment of the application proposes with reference to Fig. 3, Fig. 3.It is executing
It, can also be to user's recommendation information after complete step S11 step-step S13.As shown in figure 3, method includes the following steps:
Step S31: according to the target component, target audience is determined;
S32: Xiang Suoshu target audience of step pushes the information of the information publisher publication.
First in step S31, since target component is that information publisher is arranged Wei certain information recommendation effect is reached
Condition information publisher can be obtained for the ideal audience for the information selection recommended according to target component.Again will
It is matched by the label of step S11- step S13 user obtained with target component, can choose out ideal audient group
Body.
Such as: the target component of information publisher setting are as follows: female, 18-24 one full year of life, clicking rate 80%-100%, then in generation
The clicking rate of table user is in the user included by 80% or more label, and by gender be female, age level is 18-24 one full year of life
User be determined as target audience.
Then in step s 32, the information issued to target audience pushed information publisher.Such as: in above-mentioned steps S31
In, using affiliated age bracket be 18-24 one full year of life and to the clicking rate of recommendation information 80% or more female user as target
After audient, the message issued is needed to recommend target audience publisher.Finally, the user for belonging to target audience can view
The information of publication, and the information of publication can not be viewed by being not belonging to target audience.
In the present embodiment, the target audience for meeting target component, Ke Yibang are on the one hand selected according to the label of user
Supplementary information publisher faster and better determines target audience;On the other hand, during determining target audience, since target is joined
Several essence is to be related to various restrictive conditions, thus, the target audience chosen has great probability to recommendation
Information makes user behavior;Further, since target component can be with flexible setting, thus it can preferably meet information publisher couple
The demand of various types of target audiences.
Fig. 4 is the flow chart of the determination user that one embodiment of the application proposes estimated behavior and generate probability.Such as Fig. 4 institute
Show, step S11 the following steps are included:
Step S111: historical behavior data and outer historical behavior data of standing in the station of the user are obtained;
Step S112: being to generate generally by historical behavior data in the station of the user and outer historical behavior data input columns of standing
Rate prediction model determines that the behavior of estimating of the user generates probability.
First in step S111, since the concept of the outer historical behavior data of historical behavior data and station in standing is in step
Explanation has been made in the corresponding embodiment part S11, therefore this will not be repeated here.Wherein, in conjunction with step S21, in the station of acquisition
Historical behavior data and history in the station that outer historical behavior data of standing are after being screened according to the target component of information publisher
Behavioral data and outer historical behavior data of standing.
Then in step S112, historical behavior data and outer historical behavior data of standing can be formed in the station after screening
Every sample data input behavior is generated probability and estimates mould by a plurality of sample data about particular user behavior of multiple users
Type, the available behavior of estimating for each sample data generate probability.
In the present embodiment, during determining the label of user, with reference to multiple data sources: history in the station of user
Behavioral data and outer historical behavior data of standing.Compared to drawing a portrait in the related technology only with reference to the user of user, with reference to multiple data
Source can preferably determine the label of user.
Also, historical behavior data and outer historical behavior data of standing characterize information from different perspectives respectively in the station of user
Recommend the effect of (such as: advertising display), such as: whether user clicks the information of recommendation, whether user has purchased recommendation
Commodity in information.Historical behavior data in the station of user and outer historical behavior data of standing are combined to determine mark belonging to user
Label, obtained result more can really reflect the case where user makes feedback for the information recommended, relatively be suitable for
The scene of information recommendation can launch advertisement for advertiser and provide accuracy higher ginseng for example, being applied to launch advertising scenarios
Examine information.
In another embodiment, step S11 can with the following steps are included:
It is to generate probability prediction model by the historical behavior data input columns of the user, with estimating for the determination user
Behavior generates probability, wherein it is with multiple historical behavior sample datas for carrying label that the behavior, which generates probability prediction model,
For input, obtained model is trained to the second preset model, wherein what each historical behavior sample data carried
Whether label characterization user makes agenda.
Historical behavior sample data is marked, such as: with a kind of sign flag positive sample data, (user is for recommending
Information be made that the sample data of user behavior), with another different sign flag negative sample data, (user is for recommending
Information do not make the sample data of user behavior).Such as: it is negative with symbol " 0 " label with symbol one token positive sample data
Sample data.
Behavior, which generates probability prediction model, can multiple types, e.g. the click behavior of the probability of prediction click behavior
Probability prediction model is generated, or predicts that the collection behavior of the probability of collection behavior generates probability prediction model, or be
Predict that the behavior of the probability of other types of user behavior generates probability prediction model.Probability is generated for different types of behavior
For prediction model, the type of the historical behavior sample data needed is different, but is consistent with the type of model, such as: it is right
Probability prediction model is generated in the behavior of click, the type of the historical behavior sample data needed is the sample number about the behavior of click
According to;Probability prediction model is generated for collection behavior, the type of the historical behavior sample data needed is about collection behavior
Sample data.
Second preset model can be machine learning model, such as model-naive Bayesian, decision-tree model etc..With last
For obtained label is the label about the behavior of click, if label " 1 " indicates that user clicks the information of recommendation, label
" 0 " indicates that user does not click on the information of recommendation, then, the positive sample data of label " 1 " will be carried and carry the negative sample of label " 0 "
Notebook data is input to the second preset model and is trained, and the preset model after training has the point of prediction single user
The function that behavior generates probability is hit, the preset model after the training is that click behavior generates probability prediction model.
In the present embodiment, it pre-establishes behavior and generates probability prediction model, for each user in multiple users,
After obtaining the historical behavior data of the user, the historical behavior data of the user can be inputted to the behavior pre-established and generated generally
Rate prediction model, behavior generate the behavior of estimating that the numerical value that probability prediction model exports is the user and generate probability, thus, energy
In a fairly large number of situation of user, conveniently and efficiently determine that the behavior of estimating of user generates probability.
It is the flow chart of the method for the training calibrating patterns that one embodiment of the application proposes with reference to Fig. 5, Fig. 5.Such as Fig. 5 institute
Show, method includes the following steps:
Step S41: according to the respective size estimated behavior and generate probability of the multiple sample data, to the multiple sample
Notebook data is ranked up.
When the size for generating probability to behavior of estimating is ranked up, multiple behavior generation probability of estimating are for same tool
The user behavior of body.That is: when being directed to click behavior, multiple behavior generation probability of estimating about the behavior of click are arranged
When for collection behavior, multiple behavior generation probability of estimating about collection behavior are ranked up for sequence.Sortord, can be with
It is to carry out ascending order arrangement according to the size of probability value, is also possible to carry out descending arrangement according to the size of probability value.
Step S42: multiple sample datas after sequence are divided into multiple groups sample data.
Each is corresponding with a user about the sample data of particular user behavior, and the sample data after sequence is drawn
Corresponding user is assigned to, so that each user is corresponding with a sample data group, wherein each sample data group is equal
It include a plurality of sample data about particular user behavior of no less than preset quantity.Such as: the sample data of certain user is small
Group includes a plurality of sample data about the behavior of click.
Step S43: for every group of sample data in the multiple groups sample data, behavior is estimated to this group of sample data
It generates probability to be averaged, obtains the group and estimate behavior generation probability sample value.
All behavior generation probability of estimating of one sample data group are averaged, the user for corresponding to the group is obtained
Estimate behavior generate probability sample value.Such as: the behavior of estimating about the behavior of click of a sample data group is generated
Probability is averaged, and the behavior of estimating about the behavior of click of the corresponding user of the available group generates probability sample value.
Step S44: marking whether it is preset mark according to what sample data each in this group of sample data carried, and determining should
The label carried in group sample data is the accounting of the sample data of preset mark, and the accounting is determined as the practical row of the group
To generate probability sample value.
The a plurality of sample data about particular user behavior of each sample data group carries label, when the mark
When note is preset mark, indicate that the corresponding user of this sample data is made that corresponding user behavior.Such as: one about point
The sample data for hitting behavior carries label " 1 " (" 1 " is preset mark), indicates user to the letter of the recommendation in the sample data
Breath is made that click behavior, if a sample data about the behavior of click does not carry label " 1 ", indicates user to the sample
The information of recommendation in data does not make click behavior.The number of the sample data of preset mark will be carried in sample data group
Amount is used as molecule, and using the quantity of sample data all in sample data group as denominator, the ratio of the two is practical row
To generate probability sample value, such as: the sample data and all sample datas of carrying in above-mentioned sample data group label " 1 "
The ratio between generate probability sample value about the agenda of the behavior of click for the group corresponding user.
Step S45: behavior is estimated according to multiple groups and is generated between probability sample value and agenda generation probability sample value
Mapping relations are input, are trained to the first preset model, obtain the calibrating patterns.
Specifically, step S45 includes:
Behavior generation probability sample value and agenda generation probability sample are estimated according to the multiple sample of users is respective
This value determines that the behavior of estimating of multiple sample of users generates reflecting between probability sample value and agenda generation probability sample value
Penetrate relationship (such as: functional relation);
By the multiple sample of users estimate behavior generate probability sample value and agenda generate probability sample value it
Between mapping relations input first preset model and (can be machine learning model, such as model-naive Bayesian, decision tree
Model etc.), first preset model is trained, the calibrating patterns are obtained.
Following embodiments give the specific application process that scene is launched about advertisement.
Fig. 6 is another flow chart of the method for the acquisition historical behavior data that another embodiment of the application provides.Such as Fig. 6
It is shown, before step S11- step S13, acquire historical behavior data the step of include:
Step S10a: the marketing objectives of advertiser's setting is obtained;
Step S10b: it is respective with the associated historical behavior data of the marketing objectives to acquire the multiple user.
In the present embodiment, the marketing objectives of advertiser's setting refers to the promotion effect that advertiser demand communication reaches.Illustratively, it seeks
Pin target may is that high clicking rate, high conversion, high UV (Unique Visitor;Independent visitor) amount etc..
In view of the marketing objectives of advertiser's setting, can acquire multiple users it is respective with the marketing objectives is associated goes through
Then history behavioral data executes step S11- step S13, to obtain the use for the corresponding dimension of marketing objectives being arranged from advertiser
Family label.Illustratively, the marketing objectives of advertiser's setting is high clicking rate, execute the obtained user tag of above-mentioned steps the result is that
The result divided according to clicking rate height.
Fig. 7 is the flow chart for the method that the advertisement that another embodiment of the application provides is launched.As shown in fig. 7, in step
It is further comprising the steps of after S11- step S13:
Step S14: according to the marketing objectives, commercial audience is determined from the multiple user;
S15: Xiang Suoshu commercial audience of step pushes the marketing advertisement that the advertiser launches.
After determining commercial audience further according to the marketing objectives that advertiser is arranged, it is wide can targetedly to launch marketing
It accuses, reduces the accounting of miscarrying crowd.Specifically, according to advertiser be arranged marketing objectives, from multiple users determine advertisement by
Crowd, commercial audience is the matched user of marketing objectives being arranged with advertiser, then to the main dispensing of commercial audience advertisement
Marketing advertisement, optimize advertisement dispensing effect;To non-commercial audience, can not the main dispensing of advertisement marketing advertisement,
It reduces advertisement and launches cost.
If the marketing objectives of advertiser's setting is high clicking rate, selection represents use included by the higher label of clicking rate
Family is as commercial audience, and to the marketing advertisement of the main dispensing of commercial audience advertisement.As another example, the marketing of advertiser's setting
Target is high UV amount, then can use multiple per family as commercial audience, then to the battalion of the main dispensing of commercial audience advertisement
Sell advertisement.
In conjunction with above embodiments, another embodiment of the application provides a kind of method that advertisement is launched.In step S11- step
Outside S13, this method is further comprising the steps of:
According to the label of target user, the target user, the mesh are given in the advertisement pushing that corresponding with the label will market
Marking user is any user in the multiple user.
In the present embodiment, after determining the label of user, the orientation that can also carry out marketing advertisement is launched.Specifically
Ground can push advertisement of marketing corresponding with the label for label respectively, i.e., push different battalion to the user of different labels
Advertisement is sold, to improve the clicking rate and conversion ratio of user, optimizes advertisement delivery effect, cost is launched in control advertisement.
Based on the same inventive concept, one embodiment of the application provides a kind of device of determining label.It is this with reference to Fig. 8, Fig. 8
Apply for that an embodiment provides the schematic diagram for calibrating the device of label really.As shown in figure 8, the device includes:
First determining module 801 determines that the user's estimates behavior generation for the historical behavior data according to user
Probability;
Calibration module 802, for the behavior of estimating of the user to be generated probability input calibrating patterns, to the user
Estimate behavior generate probability calibrated, wherein the calibrating patterns be estimated with multiple groups behavior generate probability sample value and
It is input that agenda, which generates the mapping relations between probability sample value, is trained obtained mould to the first preset model
Type;
Second determining module 803 determines mark for generating probability according to the behavior of estimating after the calibration for the user
Label.
Optionally, described device further include:
First obtains module, for obtaining the target component of information publisher setting;
Acquisition module, for acquiring the user and the associated historical behavior data of the target component.
Optionally, described device further include:
Third determining module, for after the label for determining the user, according to the target component, determine target by
It is many;
Pushing module, for after the label for determining the user, Xiang Suoshu target audience to push the information publication
The information of Fang Fabu.
Optionally, described device further include:
Sorting module, for according to the multiple sample data it is respective estimate behavior generate probability size, to described
Multiple sample datas are ranked up;
Division module, for multiple sample data graduation after sequence to be divided into multiple groups sample data;
Computing module, for for every group of sample data in the multiple groups sample data, to the pre- of this group of sample data
Estimate behavior generation probability to be averaged, obtains the group and estimate behavior generation probability sample value;
4th determining module marks whether it is pre- bidding for what is carried according to sample data each in this group of sample data
Note, determines that the label carried in this group of sample data is the accounting of the sample data of preset mark, and the accounting is determined as
This group of agenda generates probability sample value.
Training module, for according to multiple groups estimate behavior generate probability sample value and agenda generate probability sample value it
Between mapping relations be input, the first preset model is trained, the calibrating patterns are obtained.
Optionally, first determining module includes:
Second obtains module, historical behavior data and outer historical behavior data of standing in the station for obtaining the user;
Input module, for being to produce by historical behavior data in the station of the user and outer historical behavior data input columns of standing
Raw probability prediction model determines that the behavior of estimating of the user generates probability.
Optionally, first determining module includes:
Input submodule, for being to generate probability prediction model by the historical behavior data input columns of the user, with true
The behavior of estimating of the fixed user generates probability, wherein it is with multiple carryings label that the behavior, which generates probability prediction model,
Historical behavior sample data is input, is trained obtained model to the second preset model, wherein each history row
Characterize whether user makes agenda for the label that data carry.
Based on the same inventive concept, another embodiment of the application provides a kind of computer readable storage medium, stores thereon
There is computer program, the side for calibrating label really as described in any of the above-described embodiment of the application is realized when which is executed by processor
Step in method.
Based on the same inventive concept, another embodiment of the application provides a kind of electronic equipment.It is the application with reference to Fig. 9, Fig. 9
The schematic diagram for the electronic equipment that one embodiment provides.As shown in figure 9, electronic equipment 100 includes: memory 110 and processor
120, it is connected between memory 110 and processor 120 by bus communication, is stored with computer program in memory 110, the meter
Calculation machine program can be run on processor 120, and then realize the method for calibrating label described in any of the above-described embodiment of the application really
In step.
For device embodiment, since it is basically similar to the method embodiment, related so being described relatively simple
Place illustrates referring to the part of embodiment of the method.
All the embodiments in this specification are described in a progressive manner, the highlights of each of the examples are with
The difference of other embodiments, the same or similar parts between the embodiments can be referred to each other.
It should be understood by those skilled in the art that, the embodiments of the present application may be provided as method, apparatus or calculating
Machine program product.Therefore, the embodiment of the present application can be used complete hardware embodiment, complete software embodiment or combine software and
The form of the embodiment of hardware aspect.Moreover, the embodiment of the present application can be used one or more wherein include computer can
With in the computer-usable storage medium (including but not limited to magnetic disk storage, CD-ROM, optical memory etc.) of program code
The form of the computer program product of implementation.
The embodiment of the present application is referring to according to the method for the embodiment of the present application, terminal device (system) and computer program
The flowchart and/or the block diagram of product describes.It should be understood that flowchart and/or the block diagram can be realized by computer program instructions
In each flow and/or block and flowchart and/or the block diagram in process and/or box combination.It can provide these
Computer program instructions are set to general purpose computer, special purpose computer, Embedded Processor or other programmable data processing terminals
Standby processor is to generate a machine, so that being held by the processor of computer or other programmable data processing terminal devices
Capable instruction generates for realizing in one or more flows of the flowchart and/or one or more blocks of the block diagram
The device of specified function.
These computer program instructions, which may also be stored in, is able to guide computer or other programmable data processing terminal devices
In computer-readable memory operate in a specific manner, so that instruction stored in the computer readable memory generates packet
The manufacture of command device is included, which realizes in one side of one or more flows of the flowchart and/or block diagram
The function of being specified in frame or multiple boxes.
These computer program instructions can also be loaded into computer or other programmable data processing terminal devices, so that
Series of operation steps are executed on computer or other programmable terminal equipments to generate computer implemented processing, thus
The instruction executed on computer or other programmable terminal equipments is provided for realizing in one or more flows of the flowchart
And/or in one or more blocks of the block diagram specify function the step of.
Although preferred embodiments of the embodiments of the present application have been described, once a person skilled in the art knows bases
This creative concept, then additional changes and modifications can be made to these embodiments.So the following claims are intended to be interpreted as
Including preferred embodiment and all change and modification within the scope of the embodiments of the present application.
Finally, it is to be noted that, herein, relational terms such as first and second and the like be used merely to by
One entity or operation are distinguished with another entity or operation, without necessarily requiring or implying these entities or operation
Between there are any actual relationship or orders.Moreover, the terms "include", "comprise" or its any other variant meaning
Covering non-exclusive inclusion, so that process, method, article or terminal device including a series of elements not only wrap
Those elements are included, but also including other elements that are not explicitly listed, or further includes for this process, method, article
Or the element that terminal device is intrinsic.In the absence of more restrictions, being wanted by what sentence "including a ..." limited
Element, it is not excluded that there is also other identical elements in process, method, article or the terminal device for including the element.
Above to method, apparatus, storage medium and the electronic equipment of a kind of determining label provided herein, carry out
It is discussed in detail, specific examples are used herein to illustrate the principle and implementation manner of the present application, above embodiments
Illustrate to be merely used to help understand the present processes and its core concept;At the same time, for those skilled in the art, according to
According to the thought of the application, there will be changes in the specific implementation manner and application range, in conclusion the content of the present specification
It should not be construed as the limitation to the application.
Claims (10)
1. a kind of method of determining label, which is characterized in that the described method includes:
According to the historical behavior data of user, determine that the behavior of estimating of the user generates probability;
By the user estimate behavior generate probability input calibrating patterns, with to the user estimate behavior generate probability into
Row calibration, wherein the calibrating patterns are to estimate behavior with multiple groups to generate probability sample value and agenda generation probability sample
Mapping relations between value are input, are trained obtained model to the first preset model;
Probability is generated according to the behavior of estimating after the calibration, determines the label of the user.
2. the method according to claim 1, wherein the method also includes:
Obtain the target component of information publisher setting;
Acquire the user with the associated historical behavior data of the target component.
3. according to the method described in claim 3, it is characterized in that, the method is also after the label for determining the user
Include:
According to the target component, target audience is determined;
The information that the information publisher issues is pushed to the target audience.
4. the method according to claim 1, wherein the calibrating patterns obtain according to the following steps:
According to the respective size estimated behavior and generate probability of multiple sample datas, the multiple sample data is ranked up;
Multiple sample datas after sequence are divided into multiple groups sample data;
For every group of sample data in the multiple groups sample data, probability progress is generated to the behavior of estimating of this group of sample data
It is average, it obtains the group and estimates behavior generation probability sample value;
Mark whether it is preset mark according to what sample data each in this group of sample data carried, determines in this group of sample data
The label of carrying is the accounting of the sample data of preset mark, and the accounting is determined as this group of agenda and generates probability sample
This value;
It is defeated for estimating behavior to generate the mapping relations between probability sample value and agenda generation probability sample value according to multiple groups
Enter, the first preset model is trained, obtains the calibrating patterns.
5. method according to claim 1 to 4, which is characterized in that according to the historical behavior data of user, determine institute
The behavior of estimating for stating user generates probability, comprising:
Obtain historical behavior data and outer historical behavior data of standing in the station of the user;
It is to generate probability prediction model by historical behavior data in the station of the user and outer historical behavior data input columns of standing, really
The behavior of estimating of the fixed user generates probability.
6. according to the method described in claim 5, it is characterized in that, determining the user according to the historical behavior data of user
Estimate behavior generate probability, comprising:
It is to generate probability prediction model by the historical behavior data input columns of the user, behavior is estimated with the determination user
Generate probability, wherein it is defeated that the behavior generation probability prediction model, which is with multiple historical behavior sample datas for carrying label,
Enter, obtained model is trained to the second preset model, wherein the label that each historical behavior sample data carries
Whether characterization user makes agenda.
7. a kind of device of determining label, which is characterized in that described device includes:
First determining module determines that the behavior of estimating of the user generates probability for the historical behavior data according to user;
Calibration module, for the behavior of estimating of the user to be generated probability input calibrating patterns, to be estimated to the user
Behavior generates probability and is calibrated, wherein the calibrating patterns are to estimate behavior with multiple groups to generate probability sample value and practical row
It is input to generate the mapping relations between probability sample value, obtained model is trained to the first preset model;
Second determining module determines label for generating probability according to the behavior of estimating after the calibration for the user.
8. device according to claim 7, which is characterized in that described device further include:
First obtains module, for obtaining the target component of information publisher setting;
Acquisition module, for acquiring the user and the associated historical behavior data of the target component.
9. a kind of computer readable storage medium, is stored thereon with computer program, which is characterized in that the program is held by processor
The step in the method as described in claim 1-6 is any is realized when row.
10. a kind of electronic equipment including memory, processor and stores the calculating that can be run on a memory and on a processor
Machine program, which is characterized in that the step of method as described in claim 1-6 is any is realized when the processor executes.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910168755.5A CN110046928A (en) | 2019-03-06 | 2019-03-06 | Determine method, apparatus, electronic equipment and the readable storage medium storing program for executing of label |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910168755.5A CN110046928A (en) | 2019-03-06 | 2019-03-06 | Determine method, apparatus, electronic equipment and the readable storage medium storing program for executing of label |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110046928A true CN110046928A (en) | 2019-07-23 |
Family
ID=67274581
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910168755.5A Pending CN110046928A (en) | 2019-03-06 | 2019-03-06 | Determine method, apparatus, electronic equipment and the readable storage medium storing program for executing of label |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110046928A (en) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110866776A (en) * | 2019-10-12 | 2020-03-06 | 上海掌门科技有限公司 | Data calibration method for popularization resource, electronic device and readable storage medium |
CN111105082A (en) * | 2019-12-05 | 2020-05-05 | 山东浪潮人工智能研究院有限公司 | Workpiece quality prediction model construction method and prediction method based on machine learning |
CN111144974A (en) * | 2019-12-04 | 2020-05-12 | 北京三快在线科技有限公司 | Information display method and device |
CN111860870A (en) * | 2020-07-29 | 2020-10-30 | 北京达佳互联信息技术有限公司 | Training method, device, equipment and medium for interactive behavior determination model |
CN112711643A (en) * | 2019-10-25 | 2021-04-27 | 北京达佳互联信息技术有限公司 | Training sample set obtaining method and device, electronic equipment and storage medium |
-
2019
- 2019-03-06 CN CN201910168755.5A patent/CN110046928A/en active Pending
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110866776A (en) * | 2019-10-12 | 2020-03-06 | 上海掌门科技有限公司 | Data calibration method for popularization resource, electronic device and readable storage medium |
CN110866776B (en) * | 2019-10-12 | 2023-11-24 | 上海掌门科技有限公司 | Data calibration method for popularization resources, electronic equipment and readable storage medium |
CN112711643A (en) * | 2019-10-25 | 2021-04-27 | 北京达佳互联信息技术有限公司 | Training sample set obtaining method and device, electronic equipment and storage medium |
CN112711643B (en) * | 2019-10-25 | 2023-10-10 | 北京达佳互联信息技术有限公司 | Training sample set acquisition method and device, electronic equipment and storage medium |
CN111144974A (en) * | 2019-12-04 | 2020-05-12 | 北京三快在线科技有限公司 | Information display method and device |
CN111105082A (en) * | 2019-12-05 | 2020-05-05 | 山东浪潮人工智能研究院有限公司 | Workpiece quality prediction model construction method and prediction method based on machine learning |
CN111860870A (en) * | 2020-07-29 | 2020-10-30 | 北京达佳互联信息技术有限公司 | Training method, device, equipment and medium for interactive behavior determination model |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110046928A (en) | Determine method, apparatus, electronic equipment and the readable storage medium storing program for executing of label | |
Daud et al. | Impact of customer trust toward loyalty: the mediating role of perceived usefulness and satisfaction. | |
CN105493057B (en) | The content selection controlled using accuracy | |
CN102880688B (en) | A kind of method for webpage is estimated, device and equipment | |
CN110097395B (en) | Directional advertisement putting method and device and computer readable storage medium | |
CN109389442A (en) | Method of Commodity Recommendation and device, storage medium and electric terminal | |
CN107341272A (en) | A kind of method for pushing, device and electronic equipment | |
CN111125574A (en) | Method and apparatus for generating information | |
CN111210258A (en) | Advertisement putting method and device, electronic equipment and readable storage medium | |
CN107767175A (en) | A kind of information based on Information rate launches processing method and processing device | |
CN113592535B (en) | Advertisement recommendation method and device, electronic equipment and storage medium | |
CN111798280B (en) | Multimedia information recommendation method, device and equipment and storage medium | |
CN115423535B (en) | Product purchasing method, device, equipment and medium based on market priori big data | |
Clottey et al. | Determining core acquisition quantities when products have long return lags | |
CN106445954A (en) | Business object display method and apparatus | |
CN112015986B (en) | Data pushing method, device, electronic equipment and computer readable storage medium | |
CN113761348A (en) | Information recommendation method and device, electronic equipment and storage medium | |
CN108829718A (en) | A kind of method and apparatus of data processing | |
CN103617146B (en) | A kind of machine learning method and device based on hardware resource consumption | |
CN109885834A (en) | A kind of prediction technique and device of age of user gender | |
CN112579910A (en) | Information processing method, information processing apparatus, storage medium, and electronic device | |
US11494686B1 (en) | Artificial intelligence system for relevance analysis of data stream items using similarity groups and attributes | |
CN114971716A (en) | Service interface quality evaluation method and device, equipment, medium and product thereof | |
CN111340522A (en) | Resource recommendation method, device, server and storage medium | |
CN114663015A (en) | Replenishment method and device |
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 |