CN104851026B - Position the primary advertisement reward system and method for bidding of user in real time based on big data - Google Patents
Position the primary advertisement reward system and method for bidding of user in real time based on big data Download PDFInfo
- Publication number
- CN104851026B CN104851026B CN201510263857.7A CN201510263857A CN104851026B CN 104851026 B CN104851026 B CN 104851026B CN 201510263857 A CN201510263857 A CN 201510263857A CN 104851026 B CN104851026 B CN 104851026B
- Authority
- CN
- China
- Prior art keywords
- user
- data
- advertisement
- user data
- center
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Abstract
The present invention provides a kind of to position the primary advertisement reward system of bidding of user behavior based on big data in real time, and data acquisition center continuous collecting initial user data on time dimension forms usable user data;Service Management Center configures the user personality and model parameter of usable user data;Data analysis center carries out corresponding user data check and correction and analysis according to the model parameter of Service Management Center in multilayer dimension, forms user data sample;Advertisement intelligent launches engine and optimizes training to user data sample by genetic algorithm, forms user data after optimization, and the ad data of data analysis center advertisement data portion is launched for user data after optimization.The bonusing method of the above-mentioned primary advertisement reward system of bidding for positioning user behavior in real time based on big data is provided simultaneously.The advertisement value influence that the present invention improves the Real time Efficiency of big data, improves advertiser, has the characteristics that Focus, directionality and interactivity.
Description
Technical field
The present invention relates to efficient mass data algorithm and big data technical fields, more specifically, being related to a kind of use
The primary prize advertisement of bidding for positioning user behavior in real time based on big data of user is quickly positioned in Internet advertising mass data
Encourage system and method.
Background technology
In Internet advertising industry, since internet data is at the growth trend of magnitude, cause data volume very huge
Greatly, and bring user behavior analysis, user modeling the problems such as be increasingly difficult to, when so as to cause user's locating query, response
Speed is slow, and query time is long, and efficiency of algorithm is low, or even goes out in the case where the excessively huge efficiency of algorithm of data volume is relatively low
Phenomena such as existing user modeling mistake, user positions failure.
And due in internet advertisement system, in order to allow advertisement to launch away in time, frequently with non-precise positioning
Algorithm is launched, therefore the quality of advertisement dispensing can not be improved, leads to the significant wastage of advertiser resource.
Invention content
The present invention in view of the above defects of the prior art, provides one kind and positioning user's row in real time based on big data
For the primary advertisement of bidding reward system and method, system solves the problem carried out in magnanimity big data the foundation of user model with
And the problem of quickly positioning user.
To achieve the above object, the present invention is achieved by the following technical solutions:
According to an aspect of the invention, there is provided a kind of, to position bidding for user behavior in real time based on big data primary wide
Reward system is accused, including data acquisition center, Service Management Center, data analysis center and advertisement intelligent launch engine;Its
In:
The data acquisition center, the continuous collecting initial user data on time dimension, and by the initial user of acquisition
The required user model that data are configured according to Service Management Center, carries out user data modeling and user behavior confirms, is formed
User data can be used;
The Service Management Center configures the user personality and model parameter of usable user data;
The data analysis center includes user data part and advertisement data portion;When data acquisition center is by magnanimity
After usable user data is sent to data analysis center, data analysis center according to the model parameter of Service Management Center,
Corresponding user data check and correction and analysis are carried out in multilayer dimension, form user data sample;
It includes data-optimized center and delivery system that the advertisement intelligent, which launches engine, described data-optimized centrally through something lost
Propagation algorithm optimizes training to user data sample, forms user data after optimization, the delivery system will be in data analysis
The ad data of heart advertisement data portion is launched for user data after optimization.
Preferably, the user personality includes static nature and behavioral characteristics, wherein the static nature includes cell-phone number
Code, name, gender, age, city, industry, income level and/or hobby, the behavioral characteristics include time, click time
Count and/or check content type.
Preferably, the model parameter includes:
User property, including phone number, name, gender, the age, city, industry, income level, hobby, when
Between characteristic etc.;
Launch policy attribute, including region, the period, the frequency, type, operating system, network type, age of user section,
Consuming capacity etc..
Preferably, the genetic algorithm is the Multi-layered Feedforward Networks algorithm based on error propagation.
Preferably, the Multi-layered Feedforward Networks include input layer, hidden layer and output layer;The input layer, hidden layer and
Output layer is composed of multiple units respectively.
Preferably, in Multi-layered Feedforward Networks, user data sample training process is being divided into information just by the genetic algorithm
To communication process and error back propagation process;Wherein:
Described information forward-propagating process is specially:Input sample-> input layer-> hidden layer-> output layers;
The error back propagation process is specially:Output error-> hidden layer-> input layers;
The main purpose of error back propagation process is:By the way that by output error anti-pass, error distribution is owned to each layer
Unit to obtain the error signal of each layer unit, and then corrects the weights of each unit;
Wherein, symbol->Indicate output extremely.
Preferably, by genetic algorithm optimization training user's data sample, include the following steps:
Step a initializes k string, specially:Select n at random from m user characteristics of user data sample, n <
M, and the string of this n user characteristics composition is encoded, the string after coding is formed, the string after coding is in initial population
Body;
Step b is trained respectively according to the corresponding user characteristics of multiple strings obtained in step a, training time difference
It is denoted as t1、t2……、tk, the Multi-layered Feedforward Networks after training are formed, are used in combination the Multi-layered Feedforward Networks after training to be predicted, in advance
The ratio between correct number and total number of the result of survey are accuracy, and the accuracy of k string is denoted as l respectively1、l2、……lk;
Step c, including following two steps:
It selects the highest string k of accuracy and carries out follow-on duplication as the highest individual of evaluation of estimate;
It selects the highest individual of evaluation of estimate and carries out mutation operation later, according to mutation probability q in the string k after coding
Each user characteristics be replaced;
Step d, repetitive operation step b to step c, by N for iteration after, the highest individual of evaluation of estimate in N+1 generation
As optimal solution.
In the step c, follow-on duplication and mutation operation, mainly for hereditary capacity and optimization offspring's sample.
Preferably, in the step d, the rule of replacement is to randomly choose one in m-n remaining user characteristics
It carries out.
According to the second aspect of the invention, provide that a kind of to position bidding for user behavior in real time based on big data primary
The execution method of advertisement reward system, includes the following steps:
Step 1, media client initiates ad-request, and by user data delivery to data acquisition center;
Step 2, data acquisition center verifies the legitimacy of user data and returns to error message if illegal first;
If legal, user data is pushed into advertisement intelligent by data analysis center and is launched in engine, advertisement intelligent dispensing is drawn
It holds up and user data optimal screening is carried out by genetic algorithm described in any one of the above embodiments;
Step 3, user data is predicted and is determined by user model, and the delivery system running that advertisement intelligent launches engine is wide
It accuses and launches election algorithm;
Step 4, advertisement launches election algorithm and chooses the advertising listing for best suiting active user's model, then forms advertisement
A queue carries out advertisement dispensing according to the rule of setting.
Preferably, the advertisement dispensing election algorithm is specially:Policy attribute is being launched according to region, time, type etc.
After the screening of condition, the ad queue that can be launched is obtained, each advertisement in the ad queue that can be launched has dispensing weight,
In the case of weighted, the maximum advertisement of weight is elected;It is identical in weight, it is obtained according to the last time wide
It accuses, subsequent one of the advertisement is elected to launch advertisement as priority;If this time launched to launch for the first time or last
To advertisement be the last one advertisement in the ad queue that can be launched, then this time first advertisement of election be it is preferential launch it is wide
It accuses.
Preferably, the rule of the setting is set according to the desirable user property of client;Such as client A wishes that orientation is thrown
It is put into the user of 20~30 years old age bracket, and consuming capacity is medium, this rule is just:30 > ages > 20, and consume
Ability=medium.
Compared with prior art, the present invention has the advantages that:
1, the present invention improves the Real time Efficiency of big data.
2, the present invention improves the advertisement value influence of advertiser.
3, the present invention realizes the high efficiency of algorithm and framework in magnanimity big data.
4, the present invention positions user using behavioral data of the user in internet, and is thrown based on genetic algorithm optimization advertisement
Put method.
5, the present invention utilizes genetic algorithm, according to all various dimensions, periodically to the analysis in user data carry out behavior, in turn
Study predicts the behavior of user after being accumulated finally by study.
6, the present invention has the characteristics that Focus, directionality and interactivity.
Description of the drawings
Upon reading the detailed description of non-limiting embodiments with reference to the following drawings, other feature of the invention,
Objects and advantages will become more apparent upon:
Fig. 1 is present system service logic figure;
Fig. 2 is that the Multi-layered Feedforward Networks structure of genetic algorithm opens up complement.
Specific implementation mode
It elaborates below to the embodiment of the present invention:The present embodiment is carried out lower based on the technical solution of the present invention
Implement, gives detailed embodiment and specific operating process.It should be pointed out that those skilled in the art
For, without departing from the inventive concept of the premise, various modifications and improvements can be made, these belong to the guarantor of the present invention
Protect range.
Embodiment
Present embodiments provide a kind of primary advertisement reward of bidding of the efficient real-time positioning user behavior based on big data
System, including data acquisition center, Service Management Center, data analysis center and advertisement intelligent launch engine;
Wherein:
Data acquisition center can ceaselessly acquire initial user data on time dimension, acquire the initial user number come
According to the required user model that can be configured according to Service Management Center, carries out user data modeling and user behavior confirms, is formed
User data can be used;
The user personality and model parameter of usable user data are configured in Service Management Center, in case in data analysis
The heart and advertisement intelligent are launched engine and are utilized;
Data analysis center is divided into two parts, and a part is user data part, and a part is advertisement data portion;Work as number
After user data can be used to be sent to data analysis center corresponding magnanimity according to acquisition center, data analysis center can be according to industry
The model parameter of business administrative center, carries out corresponding user data check and correction and analysis in multilayer dimension, forms user data sample
This;
Advertisement intelligent launches the delivery system that engine includes efficient genetic algorithm and intelligence;Advertisement intelligent launches engine
Training optimized to user data sample by genetic algorithm, forms user data after optimization, the delivery system is by data
The ad data of analysis center's advertisement data portion is launched for user data after optimization.
Above-mentioned genetic algorithm is the Multi-layered Feedforward Networks algorithm based on error propagation, by input layer, hidden layer and output layer
Three layers of composition, as shown in Fig. 2, in figure, A is input layer, and B is hidden layer, and C is output layer, WjiBetween input layer and hidden layer
Connection weight, W1jConnection weight between hidden layer and output layer, Xn are the input of n-th of neuron of input layer (unit)
Signal, Ym are the input signal of m-th of neuron of output layer (unit), and n and m are respectively natural number.Genetic algorithm was learning
Journey is divided into two stages:Information forward-propagating and error back propagation, and utilize genetic algorithm optimization training user's data sample
This, is as follows:
Step (1) selects n at random from m user characteristics of user data sample, and to this n user characteristics group
At string encoded;For example user data sample has three gender, age and income user characteristics, by these three users spy
It levies corresponding numerical value to be grouped together with binary representation and by user characteristics respectively, the string k after coding is in initial population
Individual, it is assumed that initialize 3 strings, respectively k1, k2 and k3 altogether;
Step (2) is respectively trained according to 3 corresponding user characteristics of string in step (1), the training time be denoted as t1, t2 and
T3 is used in combination network structure after training to be predicted, the ratio between correct number and total number of the result of prediction are accuracy, this
The corresponding accuracy of 3 strings is denoted as 11,12 and 13 respectively;
Step (3), including following two steps:
It selects evaluation of estimate highest individual (accuracy highest individual) and carries out follow-on duplication;
It selects evaluation of estimate highest individual and carries out mutation operation later, according to mutation probability q in the string k after coding
Each user characteristics is replaced, and the rule of replacement is from one progress of random selection in m-n remaining user characteristics;
Step (4) repeats step (2) to step (3) and is iterated, and continues to generate executable group, by N generations
After iteration, the highest individual of evaluation of estimate is optimal solution in N+1 generation.
Above-mentioned user characteristics can be divided into two class of static nature and behavioral characteristics:Static nature includes:Phone number, name,
Gender, age, city, industry, income level, hobby;Behavioral characteristics include the time, number of clicks, check content type.
Further, the model parameter includes:User property (phone number, name, gender, the age, city, industry,
Income level, hobby, time response etc.), launch policy attribute (region, period, the frequency, type, operating system, net
Network type, age of user section, consuming capacity etc.).
According to above-mentioned all various dimensions, genetic algorithm can be periodically to the analysis in user data carry out behavior, and then learns, most
Afterwards by learning to carry out the behavior prediction to user after accumulating.
The primary advertisement reward system of bidding of efficient real-time positioning user behavior provided in this embodiment based on big data,
Its bonusing method is as shown in Figure 1;The operation flow of an ad system is shown schematically in figure, wherein including mainly use
How the request of data of family client and advertisement carry out intelligent dispensing, are as follows:
Step 1, media client initiates ad-request, and by user data delivery to data acquisition center;
Step 2, data acquisition center first goes the legitimacy of verification user data, if illegal, returns to error message;Such as
Fruit is legal, then user data is pushed to advertisement intelligent by data analysis center launches in engine, and advertisement intelligent launches engine
User data optimal screening is carried out by genetic algorithm;
Step 3, user data is predicted and is determined by user model, and the delivery system operation that advertisement intelligent launches engine is wide
It accuses and launches election algorithm;
Step 4, election algorithm is launched in advertisement to choose the advertising listing for best suiting this current user model, then will be wide
It accuses and forms a queue, advertisement dispensing is carried out according to the rule of setting;
The advertisement for best suiting this user model is finally presented to this user.
Further, the advertisement dispensing election algorithm is specially:In the screening according to conditions such as region, time, types
Afterwards, the ad queue that can be launched is obtained, each advertisement has dispensing weight, and in the case of weighted, election weight is most
Big advertisement;It is identical in weight, according to last obtained advertisement, elect subsequent one of the advertisement as priority
Launch advertisement;If be this time for the first time or last advertisement be queue the last one, this time first advertisement of election is
It is preferential to launch advertisement.
Further, the rule of the setting is set according to user property desired by client.
Specifically, such as client A wish orientation launch to 20~30 years old age bracket user, and consuming capacity be it is medium,
This rule is just:30 > ages > 20, and consuming capacity=medium.
Specific embodiments of the present invention are described above.It is to be appreciated that the invention is not limited in above-mentioned
Particular implementation, those skilled in the art can make various deformations or amendments within the scope of the claims, this not shadow
Ring the substantive content of the present invention.
Claims (5)
1. a kind of primary advertisement reward system of bidding positioning user behavior in real time based on big data, which is characterized in that including number
Engine is launched according to acquisition center, Service Management Center, data analysis center and advertisement intelligent;Wherein:
The data acquisition center, the continuous collecting initial user data on time dimension, and by the initial user data of acquisition
The required user model configured according to Service Management Center, carries out user data modeling and user behavior confirms, formation can make
Use user data;
The Service Management Center configures the user personality and model parameter of usable user data;
The data analysis center includes user data part and advertisement data portion;When data acquisition center making magnanimity
After being sent to data analysis center with user data, data analysis center is according to the model parameter of Service Management Center, in multilayer
Corresponding user data check and correction and analysis are carried out in dimension, form user data sample;
It includes data-optimized center and delivery system that the advertisement intelligent, which launches engine, described data-optimized to be calculated centrally through heredity
Method optimizes training to user data sample, forms user data after optimization, and the delivery system is wide by data analysis center
The ad data for accusing data portion is launched for user data after optimization;
The genetic algorithm is the Multi-layered Feedforward Networks algorithm based on error propagation;
The Multi-layered Feedforward Networks include input layer, hidden layer and output layer;The input layer, hidden layer and output layer respectively by
Multiple unit compositions;
In Multi-layered Feedforward Networks, the training process of user data sample is divided into information forward-propagating process by the genetic algorithm
With error back propagation process;Wherein:
Described information forward-propagating process is specially:Input sample-> input layer-> hidden layer-> output layers;
The error back propagation process is specially:Output error-> hidden layer-> input layers;
By by output error anti-pass, error distribution being given to all units in each layer, to obtain institute in each layer
There is the error signal of unit, and then corrects the weights of each unit;
Wherein, symbol-> indicates output extremely;
By genetic algorithm optimization training user's data sample, include the following steps:
Step a initializes k string, specially:It is a to select n at random from m user characteristics of user data sample, n < m, and
The string of this n user characteristics composition is encoded, the string after coding is formed, the string after coding is the individual in initial population;
Step b is trained respectively according to the corresponding user characteristics of k string obtained in step a, and the training time is denoted as t respectively1、
t2……、tk, the Multi-layered Feedforward Networks after training are formed, are used in combination the Multi-layered Feedforward Networks after training to be predicted, the result of prediction
Correct number and the ratio between total number be accuracy, the k accuracy gone here and there is denoted as 1 respectively1、12、……1k;
Step c, including following two steps:
It selects the highest string k of accuracy and carries out follow-on duplication as the highest individual of evaluation of estimate;
It selects the highest individual of evaluation of estimate and carries out mutation operation, according to mutation probability q to each in the string k after coding
User characteristics are replaced;The rule of replacement is from one progress of random selection in m-n remaining user characteristics;
Step d, repetitive operation step b to step c, by N for iteration after, the highest individual of evaluation of estimate is in N+1 generation
Optimal solution.
2. the primary advertisement reward system of bidding according to claim 1 for positioning user behavior in real time based on big data,
It is characterized in that, the user personality includes static nature and behavioral characteristics, wherein the static nature includes phone number, surname
Name, gender, age, city, industry, income level and/or hobby, the behavioral characteristics include the time, number of clicks and/
Or check content type.
3. the primary advertisement reward system of bidding according to claim 1 for positioning user behavior in real time based on big data,
It is characterized in that, the model parameter includes:
User property, including phone number, name, gender, the age, city, industry, income level, hobby and/or when
Between characteristic;
Launch policy attribute, including region, the period, the frequency, type, operating system, network type, age of user section and/or
Consuming capacity.
4. a kind of execution method for the primary advertisement reward system of bidding being positioned user behavior in real time based on big data, feature are existed
In including the following steps:
Step 1, media client initiates ad-request, and by user data delivery to data acquisition center;
Step 2, data acquisition center verifies the legitimacy of user data and returns to error message if illegal first;If
It is legal, then user data is pushed into advertisement intelligent by data analysis center and launched in engine, it is logical that advertisement intelligent launches engine
It crosses the genetic algorithm described in any one of claims 1 to 3 and carries out user data optimal screening;
Step 3, user data is predicted and is determined by user model, and the delivery system running advertisement that advertisement intelligent launches engine is thrown
Put election algorithm;
Step 4, advertisement launches election algorithm and chooses the advertising listing for best suiting active user's model, and advertisement is then formed portion
Queue carries out advertisement dispensing according to the rule of setting;
Election algorithm is launched in the advertisement:After according to the screening for launching policy attribute condition, obtain to launch wide
Accuse queue, wherein each advertisement in the ad queue that can be launched has dispensing weight:
In the case of weighted, the maximum advertisement of weight is elected;
Identical in weight, according to last obtained advertisement, the last obtained advertisement of election is subsequent one wide
Accuse the advertisement launched as this second priority;If this is to launch advertisement or last obtained advertisement for the first time for that can launch
Ad queue in the last one advertisement, then this time first advertisement in the ad queue that can be launched is elected to throw as priority
Put advertisement.
5. the primary advertisement reward system of bidding according to claim 4 for being positioned user behavior in real time based on big data is held
Row method, which is characterized in that the rule of the setting is set according to the desirable user property of client.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510263857.7A CN104851026B (en) | 2015-05-21 | 2015-05-21 | Position the primary advertisement reward system and method for bidding of user in real time based on big data |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510263857.7A CN104851026B (en) | 2015-05-21 | 2015-05-21 | Position the primary advertisement reward system and method for bidding of user in real time based on big data |
Publications (2)
Publication Number | Publication Date |
---|---|
CN104851026A CN104851026A (en) | 2015-08-19 |
CN104851026B true CN104851026B (en) | 2018-07-17 |
Family
ID=53850655
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510263857.7A Active CN104851026B (en) | 2015-05-21 | 2015-05-21 | Position the primary advertisement reward system and method for bidding of user in real time based on big data |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN104851026B (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112152741A (en) * | 2019-06-28 | 2020-12-29 | 华为技术有限公司 | Channel model training method and device |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106600369A (en) * | 2016-12-09 | 2017-04-26 | 广东奡风科技股份有限公司 | Real-time recommendation system and method of financial products of banks based on Naive Bayesian classification |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102033883A (en) * | 2009-09-29 | 2011-04-27 | 阿里巴巴集团控股有限公司 | Method, device and system for improving data transmission speed of website |
CN102236867A (en) * | 2011-08-15 | 2011-11-09 | 悠易互通(北京)广告有限公司 | Cloud computing-based audience behavioral analysis advertisement targeting system |
CN103617230A (en) * | 2013-11-26 | 2014-03-05 | 中国科学院深圳先进技术研究院 | Method and system for advertisement recommendation based microblog |
-
2015
- 2015-05-21 CN CN201510263857.7A patent/CN104851026B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102033883A (en) * | 2009-09-29 | 2011-04-27 | 阿里巴巴集团控股有限公司 | Method, device and system for improving data transmission speed of website |
CN102236867A (en) * | 2011-08-15 | 2011-11-09 | 悠易互通(北京)广告有限公司 | Cloud computing-based audience behavioral analysis advertisement targeting system |
CN103617230A (en) * | 2013-11-26 | 2014-03-05 | 中国科学院深圳先进技术研究院 | Method and system for advertisement recommendation based microblog |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112152741A (en) * | 2019-06-28 | 2020-12-29 | 华为技术有限公司 | Channel model training method and device |
CN112152741B (en) * | 2019-06-28 | 2021-11-19 | 华为技术有限公司 | Channel model training method and device |
Also Published As
Publication number | Publication date |
---|---|
CN104851026A (en) | 2015-08-19 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111966698B (en) | Block chain-based trusted federation learning method, system, device and medium | |
Thurman et al. | Persist in place or shift in space? Evaluating the adaptive capacity of species to climate change | |
CN110647765A (en) | Privacy protection method and system based on knowledge migration under collaborative learning framework | |
CN110033314A (en) | Advertisement data processing method and device | |
Eklöf et al. | Secondary extinctions in food webs: a Bayesian network approach | |
CN104281882A (en) | Method and system for predicting social network information popularity on basis of user characteristics | |
CN106294830A (en) | The recommendation method and device of multimedia resource | |
CN102075352A (en) | Method and device for predicting network user behavior | |
CN110321291A (en) | Test cases intelligent extraction system and method | |
CN105760649B (en) | A kind of credible measure towards big data | |
CN104851026B (en) | Position the primary advertisement reward system and method for bidding of user in real time based on big data | |
CN108595526B (en) | Resource recommendation method and device | |
CN108305181A (en) | The determination of social influence power, information distribution method and device, equipment and storage medium | |
CN110533437A (en) | Budget allocation method and device is launched in advertisement | |
CN116523293A (en) | User risk assessment method based on fusion behavior flow chart characteristics | |
US20220343243A1 (en) | Systems and methods for resource allocation in ride-hailing platforms | |
Estrada et al. | Assessment of the National Park network of mainland Spain by the Insecurity Index of vertebrate species | |
JP2004265190A (en) | Learning method of hierarchical neutral network, its program, and recording medium having the program recorded thereon | |
Khudhair | An empirical test of information system success model in a University's electronic services | |
CN114528992A (en) | Block chain-based e-commerce business analysis model training method | |
Gupta et al. | A hybrid biogeography based heuristic for the mirrored traveling tournament problem | |
Cotta et al. | Off-policy evaluation of probabilistic identity data in lookalike modeling | |
CN116226540B (en) | End-to-end federation personalized recommendation method and system based on user interest domain | |
Chung et al. | Can theories be tested? A cryptographic treatment of forecast testing | |
KR102095010B1 (en) | Method for managing player of game |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
EXSB | Decision made by sipo to initiate substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant | ||
TR01 | Transfer of patent right | ||
TR01 | Transfer of patent right |
Effective date of registration: 20180929 Address after: 201824 Shanghai Jiading District real new street, Jinsha River road 3131, 1, 1, 1165 rooms. Patentee after: Shanghai Bingku network Polytron Technologies Inc. Address before: 200051 217G 668, Shanghai Road, Baoshan District, Shanghai. Patentee before: SHANGHAI BINGU NETWORK TECHNOLOGY Co.,Ltd. |
|
PP01 | Preservation of patent right | ||
PP01 | Preservation of patent right |
Effective date of registration: 20221108 Granted publication date: 20180717 |