CN110222267A - A kind of gaming platform information-pushing method, system, storage medium and equipment - Google Patents
A kind of gaming platform information-pushing method, system, storage medium and equipment Download PDFInfo
- Publication number
- CN110222267A CN110222267A CN201910491711.6A CN201910491711A CN110222267A CN 110222267 A CN110222267 A CN 110222267A CN 201910491711 A CN201910491711 A CN 201910491711A CN 110222267 A CN110222267 A CN 110222267A
- Authority
- CN
- China
- Prior art keywords
- user
- label
- information
- game
- consumption
- 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.)
- Granted
Links
Classifications
-
- A—HUMAN NECESSITIES
- A63—SPORTS; GAMES; AMUSEMENTS
- A63F—CARD, BOARD, OR ROULETTE GAMES; INDOOR GAMES USING SMALL MOVING PLAYING BODIES; VIDEO GAMES; GAMES NOT OTHERWISE PROVIDED FOR
- A63F13/00—Video games, i.e. games using an electronically generated display having two or more dimensions
- A63F13/70—Game security or game management aspects
- A63F13/79—Game security or game management aspects involving player-related data, e.g. identities, accounts, preferences or play histories
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9535—Search customisation based on user profiles and personalisation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/958—Organisation or management of web site content, e.g. publishing, maintaining pages or automatic linking
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Landscapes
- Engineering & Computer Science (AREA)
- Databases & Information Systems (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Business, Economics & Management (AREA)
- Computer Security & Cryptography (AREA)
- General Business, Economics & Management (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The present invention relates to gaming platform information-pushing method, system, storage medium and equipment, by carrying out data analysis to mass users data in server and establishing corresponding label, can help operation personnel quickly, accurately position user, it is comprehensive, grasp user characteristics with multi-angle.Compared with the existing technology, the present invention will realize the Rapid matching between push content and user demand, and improve matched accuracy, save Internet resources by carrying out point group to user and according to allocated user group progress information push.
Description
Technical field
The present invention relates to game to run field, is situated between more particularly, to a kind of gaming platform information-pushing method, system, storage
Matter and equipment.
Background technique
With the fast development of Internet technology, game industry also makes rapid progress, and various types of games emerges one after another, game operation
Competition between quotient is also more and more fierce, and in the case where game experiencing difference is little, stable operation is determined with good service
Can determine game show one's talent.And it is higher and higher in the cost for currently obtaining new user, customer churn is also in game operation
Common phenomenon, in order to maintain the quantity of user, gaming platform often carries out the push of game information to user to guarantee user
Viscosity.
However, gaming platform is often directed to total user when carrying out information push to user and carries out in practical business
Information push, is easy to appear pushed information and the incongruent situation of user's actual need, and excessive unrelated pushed information occupies net
Due to mismatching with user demand while network resource, lack the attraction to user, so that pushed information effect is poor.
Summary of the invention
The purpose of the present invention is to overcome the shortcomings of the existing technology with it is insufficient, a kind of push content and user demand are provided
Match, gaming platform information-pushing method, system, storage medium and the equipment that matching is high.
A kind of gaming platform information-pushing method, comprising the following steps:
Obtain the user data in server；
User base information label, customer consumption prediction label, game preference label and loss are established based on user data
User tag；
Point group is carried out to user according to user data and label and information push is carried out according to allocated user group.
Compared with the existing technology, this case is by carrying out data analysis to mass users data in server and establishing corresponding
Label, can help operation personnel quickly, accurately position user, it is comprehensive, grasp user characteristics with multi-angle, by with
Family carries out point group and carries out information push according to allocated user group, improves matched accuracy, saves network money
Source.
Further, described the step of establishing user base information label based on user data, includes:
User's registration time, registration time length are obtained, the information such as registration equipment establish registration behavior label；
Counting user logs in IP and parses user's entry address, establishes city level label where user；
Counting user login time point calculates user's active period, establishes user and enliven type label；
According to user's login time, user's active index is calculated, user's active index label is established；
Movable consumption number of times, spending amount are participated according to user, user browsing behavior is analyzed, establishes consumer consumption behavior
Label and customer consumption index label.By establishing user base information label, facilitate it is comprehensive, user is carried out with multi-angle
The Rapid matching between push content and user preferences is realized in analysis.
Further, described the step of establishing customer consumption prediction label based on user data, includes:
User data is acquired from server log；
Extract paying customer and user base attributive character and game behavior property spy without paying customer in a cycle
Sign, and Feature Engineering processing is carried out to the user base attributive character and game behavior property feature；
It promotes decision tree building disaggregated model using gradient and cross-validation method is rolled over by K and adjust disaggregated model parameter；
The user base attributive character and game behavior property feature of paying customer in a cycle are extracted, and to the use
Family primary attribute feature and game behavior property feature carry out Feature Engineering processing；
It is promoted to return building regression model and roll over cross-validation method by K using gradient and adjusts Parameters in Regression Model；
Formulate customer consumption hierarchy rules；
The disaggregated model, regression model and customer consumption hierarchy rules are integrated into customer consumption prediction model.It uses
Gradient, which is promoted, returns building regression model, and all types of data, including discrete type feature and continuous type feature can be setup flexibly,
Stronger to the robustness of exceptional value, accuracy rate is higher.
Further, described the step of establishing game preference label based on user data, includes:
Counting user generates corelation behaviour event number before registering each type game, and calculates acquisition corresponding types with this
The initial degree of liking of game；
Time that counting user puts into each type game in the period and in the time of investment of the upper period, and with
This calculates the growth rate of each type playtime increment；
User's degree of liking of every kind of type of play is calculated according to following manner:
Wherein, αiFor the growth rate of user's playtime increment, Δ x is the playtime of user's investment, TjIndicate user
The initial value for degree of liking, ti-tjIndicate time interval, M indicates consumption total value of the user in certain class game, and λ is constant, is indicated
The ratio that spending limit increases temperature, TiIndicate user in tiDegree of liking of the period to certain class game.By to building user's trip
Play preference label facilitates subsequent operation personnel to be directed to specified crowd and implements specific marketing strategy.
Further, described the step of establishing loss user tag based on user data, includes:
User data and label are associated with using user's unique identification, obtain characteristic data set；
Feature selecting is carried out to characteristic data set using gradient boosted tree and feature weight；
If using the trained butt learning model of K folding cross-validation method, and if according to the output result structure of butt learning model
Build Fusion Model；
It calls Fusion Model identification to be lost user and generate and is lost user tag information.By utilizing a variety of single models
The various behavioral characteristics of different user are practised, Fusion Model identification is called to be lost user, the prediction for improving convection current appraxia family is quasi-
True property.
Further, further comprising the steps of: by user base information label, customer consumption prediction label, game preference
Label and the bookmark name of user tag is lost as key, the user under label is converted to the corresponding value of the key, by the key and value
It is stored in a manner of mutual corresponding into database；Wherein, which specifically includes:
Obtain the unique identification of bookmark name, maximum user identifier and all users with the label；
The unique identification of users all in label is converted to by bitmap algorithm to k bitmap arrays；Wherein, k=1+
N/32, N are user's number of label；
Bitmap array is converted into hexadecimal string, and using bookmark name as key, the hexadecimal string
As the corresponding value of the key, storage is in the database.Label information is stored using bitmap algorithm, not only reduces use
Memory headroom, also greatly increase query performance.
Further, it is described according to user data and label to user carry out point group and according to allocated user group into
The step of row information pushes specifically includes:
It is higher than given threshold for consumption and consumes the push for carrying out coupon information lower than the user of given threshold, and supervises
Control its consumption；
Call back message push is carried out for loss user and the user that will be lost, and monitors it and recalls number of users；
The variation of monitoring information push front and back number of users is simultaneously shown.It is realized by the user to different labels different
The push of information and the monitoring for carrying out number of users variation intuitively illustrate the information push effect of gaming platform, are subsequent
Activity provides data and supports.
The present invention also provides a kind of gaming platform information transmission systems, comprising:
Data acquisition module, for obtaining the user data in server；
Label establishes module, for establishing user base information label, customer consumption prediction label, trip based on user data
Play preference label and loss user tag；
Info push module, for carrying out point group to user according to user data and label and according to allocated user group
Body carries out information push.
Compared with the existing technology, the module that this case provides whole set of system cooperates jointly, by extra large in server
Amount user data carry out data analysis simultaneously establish corresponding label, can help operation personnel quickly, accurately position user, entirely
User characteristics are grasped in orientation with multi-angle, are realized the Rapid matching between push content and user preferences, are improved matched standard
True property, saves Internet resources.
The present invention also provides a kind of computer readable storage mediums, store computer program thereon, it is characterised in that:
The computer program realizes the step of above-mentioned gaming platform information-pushing method when being executed by processor.
The present invention also provides a kind of computer equipment, including reservoir, processor and it is stored in the reservoir
And the computer program that can be executed by the processor, the processor realize above-mentioned game when executing the computer program
The step of platform information method for pushing.
In order to better understand and implement, the invention will now be described in detail with reference to the accompanying drawings.
Detailed description of the invention
Fig. 1 is the flow chart of gaming platform information-pushing method described in the embodiment of the present invention 1；
Fig. 2 is gaming platform information-pushing method architecture diagram described in embodiment 1 in the present invention；
Fig. 3 is that K folding cross validation building Fusion Model flow chart is utilized in the embodiment of the present invention 1；
Fig. 4 is the structure chart of Fusion Model in the embodiment of the present invention 1；
Fig. 5 is the structure chart of gaming platform information transmission system described in the embodiment of the present invention 2.
Specific embodiment
Embodiment 1
It is the flow chart of gaming platform information-pushing method of the present invention referring to FIG. 1-2,.
A kind of gaming platform information-pushing method, comprising the following steps:
S1: the user data in server is obtained；
In user data step in the acquisition server, the user data in the server refers to that multiple product reports
To the data (including but not limited to own platform product data, third party's data of crawler acquisition) of server, pass through ETL
(Extract-Transform-Load) user's entirety data set that technology obtains, specifically includes:
The structuring that reports in server database, unstructured data and report to server local Disk Logs text
Part；Wherein, the database includes but is not limited to MySQL, MongoDB, HBase database, described to report to server data
Structuring, unstructured data in library include user's registration table, user logs in behavior table, user's order table, user supplement table with money,
User browses detail list etc., specifically include user's unique identification, age of user, registion time, login time, spending amount,
The structural datas such as discount coupon consumption details；It is described to report to server local Disk Logs file, including produced when follow-up user
The unstructured datas such as raw journal file.Wherein, for unstructured data, will be corresponded in system the log of user's generation into
Row combination, is spliced into the text-independent of corresponding user's unique identifier, and cleaned using participle tool.Specifically,
Word segmentation processing is carried out using participle tool and removes stop words, obtains corpus, then user's key is obtained by TF-IDF algorithm
Word, the TF-IDF algorithm are usually used in obtaining the significance level of word in file, and calculating process is not done in the present embodiment to be had
Body limits.Participle tool described in this step includes but is not limited to the participles tool such as CoreNLP, Jieba.
S2: user base information label, customer consumption prediction label, game preference label and stream are established based on user data
Lose user tag.In other embodiments, user can also increase according to actual needs or re-establish other users in this step
Label information is convenient for subsequent exploration more users label information, plays the value of data-driven to the maximum extent.
Wherein, data analysis is carried out for user data, the user base information labelling step of establishing includes:
S211: obtaining user's registration time, registration time length, and the information such as registration equipment establish registration behavior label；
S212: counting user logs in IP and parses user's entry address, establishes city level label where user；
S213: counting user login time point calculates user's active period, establishes user and enliven type label；Wherein, institute
Stating and enlivening type includes active type or night active type in the daytime.
S214: according to user's login time, user's active index is calculated, establishes user's active index label, wherein user
Active index temporally granularity division, such as nearly T days login number of days, nearly T days offline number of days etc. can carry out overall merit, specifically, institute
User's active index is stated to obtain by following manner:
Wherein, ActiveIndex is user's active index, wiFor the weighted value for enlivening index, T is User Activity period, Xi
Index is enlivened in cycle T day for user, N is the evaluation event number for enlivening index；For example, in the present embodiment, N=3,
X1For number of days offline in the period, X2To log in number of days, X in the period3Log in for last time log date in the period and for the first time the date
Difference.
Measure user liveness index label are as follows:
Wherein, ActiveIndexLabel is liveness index, indicates that the user is low active when liveness index is 1,
Liveness index indicates that the user is generally active when being 2, and liveness index indicates user's high activity for 3.
S215: participating in movable consumption number of times, spending amount according to user, in conjunction with user browsing behavior data, establishes and uses
Family consumer behavior label and customer consumption index label.
Wherein, consumer consumption behavior label is established in S215 step includes:
The ratio that movable consumption number of times Yu overall consumption number are participated in conjunction with user establishes user to Sensitivity Index living of marketing
Label；
Do qualitative analysis in conjunction with reasonable consumption index of the user browsing behavior data to user, for example, analysis user click into
Number, browsing equipment quantity and its duration relationship for browsing to purchase equipment for the first time for entering the details page of purchase equipment, by user
Qualitative is impulsion consumption-orientation, reasonable consumption type, hesitation consumption-orientation etc., for example, when user first enters equipment details page and does not have
It removes to browse other equipment details pages, just buys this equipment, then its qualitative established user's rationality for modes such as impulsion consumption-orientations and disappear
Take index label.
Wherein, in S215 step, the step of establishing customer consumption index label, includes:
Count the consumption frequency, the last consumption time, the consumption total amount, maximum spending amount, special price of all users
The frequency, the high price consumption frequency are consumed as data set；
The quartile for calculating every a kind of consumption attribute in data set, according to the size relation of attribute value and each quartile
It scores customer consumption attribute；The scoring item includes that the consumption frequency scores, the last consumption time scores, consumption is total
Amount of money scoring, maximum spending amount scoring, the scoring of the preferential consumption frequency, high price consumption frequency scoring.
Customer consumption index is obtained to each attribute ratings item weighted calculation,
Wherein, wiThe weight of respectively each item rating, i are scoring item number, i ∈ [1,2,3,4,5,6].
According to customer consumption index, customer consumption index label is established.
Specifically, the step of carrying out data analysis for user data in step S2, establishing customer consumption prediction label packet
It includes:
S221: it is acquired from server log and pre-processes user data；Wherein, the pretreatment user data step packet
It includes:
Association user device number and user account, filtration correlation unit exception user and the multiple user accounts of filtration correlation
Device number, such user may practice in such as generation for improper user.
Exception Filter paying customer；
Ignore the attribute that attribute value largely lacks；
One-hot coding is carried out to discrete features such as channel type, cell phone platforms；
Quantify user's game events；
By the user bases attribute dualization such as cell-phone number, identity card.
S222: paying customer and user base attributive character and game behavior category without paying customer in a cycle are extracted
Property feature, and Feature Engineering processing is carried out to the user base attributive character and game behavior property feature；
Wherein, the user base attribute includes: gender, age, Platform Type, registration channel, registered number of days, login
Number of days, associated account number number, VIP grade, login times, common IP area；User's game behavior property includes: user's game
Quantity, user's game grading, user's game role grade, user's game grading, game events number, participates in user's type of play
Key activities number, game duration；It is described that feature work is carried out to the user base attributive character and game behavior property feature
Journey processing step specifically includes:
Classify to game events and counts every kind of classification total number of events；Line when filtration application PCA or other method of descents are handled
The very big feature of sexual intercourse, while calculating the mean value of homogenous characteristics, intermediate value etc. as new feature and data set is added；Filter variance
For 0 or close to 0 attribute；
It scores using RF value, F value and association relationship feature, selects the intersection of the user characteristics of three kinds of scorings preceding ten
As user's disaggregated model key feature.
S223: decision tree building disaggregated model is promoted using gradient and cross-validation method adjusting parameter is rolled over by K；Specifically
Ground promotes decision tree (Gradient Boosting Decision Tree, GBDT) using gradient and establishes disaggregated model, passes through K
Folding cross-validation method obtains optimal classification model, for predicting whether user pays.
S224: the user base attribute and game behavior property feature of paying customer in a cycle are extracted, and to described
User base attribute and game behavior property feature carry out Feature Engineering processing；Specifically, user data is extracted to concentrate in cycle T
User base attribute, consumer consumption behavior attribute and the user's game behavior property of paying customer, and use S222 the method
Feature Engineering processing is carried out to the user base attributive character and game behavior property feature.Wherein consumer consumption behavior attribute
It include: consumption total value, consumption number of times, last time consumption time, last time spending amount, average consumption in the period in the period
The amount of money, the high monovalent commodity consumption frequency, maximum spending amount.
S225: it is promoted using gradient and returns building regression model and cross-validation method adjusting parameter is rolled over by K；It is excellent at one
In the embodiment of choosing, recurrence (Gradient boosting regression, GBR) is promoted using gradient and establishes regression model,
Cross-validation method is rolled over by K and obtains optimum regression model, predicts the payment amount of user.
The gradient that the present invention uses promotes disaggregated model, gradient promotes the K that regression model uses and rolls over cross-validation method, every time
K-1 are selected as training set, remaining 1 is used as test set, obtains K group training set and test set at this time.Select one group of instruction
Practice collection, train a base learner with initial weight makes according to the error rate of the base learner come the weight in more new samples
In the base learner the high training sample point weight of error rate get higher, i.e., so that the high sample of error rate learns in next base
Paid attention in device；Then next base learner is trained using the training set after adjustment weight, recycled until base learner
Quantity reaches specified quantity T；Finally by this T base learner combination, strong learner is obtained；K group training set is traversed, K group is obtained
Strong learner therefrom selects the smallest strong learner of mean square error.It is available to be setup flexibly respectively using gradient Promotion Strategy
Categorical data, including discrete type feature and continuous type feature, it is stronger to the robustness of exceptional value, while than using single model
Obtained accuracy rate is also higher.
S226: customer consumption hierarchy rules are formulated；Specifically, the user group paid in user data set cycle T is calculated
Quantile formulates 5 customer consumption sections according to quantile, defines the customer consumption grade in each section, and according to from it is small to
Big sequence is successively defined as 1 to 5 grades.
S227: above-mentioned disaggregated model, regression model and customer consumption hierarchy rules are integrated into customer consumption prediction model.
The customer consumption prediction model can be predicted whether pay in N days user's future, be the user of payment for prediction result, by returning
Return the payment amount that model is predicted, payment amount is mapped as customer consumption grade and exports.
In described the step of establishing game preference label based on user data, since user is to the preference one of certain type game
As all there is timeliness, i.e. the favorable rating of user can decline with the growth of time, and therefore, the present invention uses improved base
In the time attenuation function model of Newton's law of cooling (Newton's law of cooling), and calculate with this trip of user
Play preference label, specifically, comprising the following steps:
S231: counting user generates corelation behaviour event number before registering each type game, wherein the corelation behaviour
Include: search game, click game introduction, forum, evaluation etc. in platform；One relevant logout of every generation, then correspond to
Initial degree of the liking T of type of playjIncrease a unit value；
S232: time that counting user puts into each type game in the period and in investment of the upper period
Between, and calculate with this growth rate of each type playtime increment；If user likes playing in the recent period the game of some type,
It can then devote considerable time, frequent operation and behavior are abundant；If user loses interest to the game, game is often gradually decreased
Time and operation behavior.Therefore, the growth rate α of user's playtime incrementiIt may be expressed as:
Wherein, tjFor a certain period, tiFor another Game Cycle, αiFor the growth rate of user's playtime increment, Δ Xj
It is user in period tjThe playtime of investment, Δ XiIt is user in period tiThe playtime of investment.
If αiIt is positive, indicates that the degree of liking of user rises；Otherwise decline.
S233: user's degree of liking T of every kind of type of play is calculated according to following manneriValue:
Wherein, αiFor the growth rate of user's playtime increment, Δ x is the playtime of user's investment, TjFor user's happiness
The initial value of love degree, ti-tjIndicate time interval, M indicates consumption total value of the user in certain class game, and λ is constant, indicates to disappear
Take the ratio that amount increases temperature, TiIndicate user in tiDegree of liking of the period to certain class game.
S234: being ranked up according to degree of liking of the user to type of play each in the period, uses according to liking that degree value determines
The type of play of family preference.Wherein, it for the computation rule of the initial degree of liking of user, can be formulated according to specific application scenarios；
For the same batch type of play being not much different by end value after initially metric being liked then to calculate, subsequent cycle can be by direct
Comparing T is worth sequence of the user to same batch type of play favorable rating.
Establishing the step of being lost user tag based on user data includes:
S241: user data and label are associated with using user's unique identification, obtain characteristic data set；Specifically, comprehensive
Above-mentioned user data set and label data are associated with to obtain characteristic data set by user's unique identification.Wherein, user data set
Including but not limited to user's registration age, registration IP, nearly N days login times, nearly N days comments number, nearly N days like times etc.
Field；Module is established based on label, label data includes but is not limited to the city level of user's login, customer consumption type, disappears
Take the fields such as index, game favorable rating value.
S242: feature selecting is carried out to characteristic data set using gradient boosted tree and feature weight；
In one embodiment, described that feature selecting is carried out to characteristic data set using gradient boosted tree and feature weight
Before step, feature pretreatment is carried out to characteristic data set, step includes: that default value is abandoned or filling is handled, for losing
The big sample of default value accounting is abandoned, for qualitative features by 0 filling, quantitative characteristic value then presses the filling of its mean value；For different dimensions, press
Nondimensionalization processing guarantees that it is in same specification, facilitates and compares.The nondimensionalization can be using such as standardization, section scaling
Etc. data nondimensionalization processing mode common in the art；For information redundancy, by it includes effective information press section
Divide, such as nearly N days online hours, be only concerned whether online hours reach some threshold value, be processed into 0 and 1 be used for indicate not
Up to threshold value and threshold value is reached；
In described the step of carrying out feature selecting using gradient boosted tree and feature weight, the feature selecting based on tree is utilized
Algorithm guarantee accuracy it is optimal in the case where, feature is selected further according to feature weight, wherein selecting according to feature weight
It selects and is characterized in investigating feature and target value correlation, using the method training of machine learning, obtain the weight coefficient of each feature,
Remove the lower feature of weight coefficient.In other embodiments, filtration method (Filter) can also be used in this step, special by investigating
Whether sign dissipates and is scored according to its diversity each feature, and the quantity of given threshold or threshold value to be selected carries out special
Sign selection.
S243: if using the trained butt learning model of K folding cross-validation method, and if according to the output knot of butt learning model
Fruit constructs Fusion Model；Specifically, using the predictive ability for improving model based on the fusion method of Stacking, compare list
One prediction model, the different user that the advantage of Fusion Model is that it can be arrived in conjunction with a variety of single model learnings are various
Behavioral characteristic, can also embody good robustness in a variety of contexts, construct Fusion Model in this step the following steps are included:
Feature selecting and the pretreated data set of feature are divided into N equal portions, wherein each base learning model passes through training
N-1 parts of data sets, remaining 1 part is used as test set, using the prediction result of all base learning models as training set and as under
The input of one step.Wherein base learning model includes but is not limited to XGBoost, random forest (Random Forest, RF), logic
Return (Logistic Regression, LR) and neural network (Neural Network, NN).In a preferred embodiment
In, the test set and training set of different models are distinguished, using the method for K folding cross validation in training base learning model to increase
Difference between model improves the effect of Model Fusion, specifically includes: above-mentioned N-1 parts of training set is traversed, to wherein traversing
Every a training set, be split as K equal portions, select K-1 every time as training set, remaining 1 part is used as test set,
Obtain K group training set and test set；The different base learning model instruction of K kind is respectively adopted to above-mentioned K group training set and test set
Practice, such as XGBoost, RF, LR, as shown in Figure 3；K base learning model is trained into the prediction result of model as the second layer
The input of Model Fusion.Wherein, the K kind difference base learning model is in combination with actual prediction as a result, discard portion is ineffective
Model.
S244: it calls Fusion Model identification to be lost user and generate and is lost user tag information.
The customer churn probability value that base learning model is exported is as the input of Fusion Model.As shown in figure 4, in fused layer
XGBoost and LR is respectively adopted and is respectively trained and predicts customer churn probability, in one embodiment, to two model outputs
Probability value take final output probability of the mean value as fused layer.In other embodiments, to the probability of two models output
Final output probability of the value using average weighted mode as fused layer.When mathematical expectation of probability is greater than the threshold value of setting, then sentence
The fixed user is to be lost user.
It inputs data set to be predicted and the Fusion Model is called to be calculated, output customer churn list is as loss user
Label information.
In one embodiment, the gaming platform information-pushing method is further comprising the steps of: using bookmark name as
Key, the mode that the user under the label is converted to the corresponding value of the key are pre- by the user base information label, customer consumption
Mark label, game preference label and loss user tag are stored into database；By that will be marked using bitmap algorithm in this step
It signs information storage in the database, i.e., marks the corresponding value of some element (Value) using one bit, key (Key) is made
For the element itself, comprising the following steps:
The bookmark name Lable_k, the maximum user identifier M that are obtained according to label model and the institute with the label are useful
Family:
<uid_1,uid_2,…,uid_n>；
Wherein, all user's unique identifications are mapped as the integer numerical value from increasing sequence, uid_1, uid_2 ..., uid_n
For user's unique identification of all users under the label；
It will<uid_1,uid_2,…,uid_n>by bitmap algorithm, it is converted to bitmap array L, wherein L are as follows:
<b_uid_1,b_uid_2,…,b_uid_k>；
Wherein, k=1+N/32, k are bitmap array number.
Bitmap array L is converted to hexadecimal string, and using bookmark name Lable_k as key (Key), it is above-mentioned
Hexadecimal string is as the corresponding value (Value) of key, and storage is in the database.Wherein, the database includes but unlimited
In Redis, MySQL, Mongo database.
Under normal circumstances, the storage of an integer needs to occupy 4 bytes, i.e., 32 bit, the storage of N number of integer, then
The memory headroom of 32N bit need to be occupied, and inquires some integer and whether there is, N number of integer need to be traversed to judge, the time is complicated
Degree is O (n)；Storage mode described in the present embodiment indicates an integer by using one bit, and the storage of N number of integer is only
The memory headroom of N number of bit need to be occupied, memory becomes original 1/32, also, for Redis database, some is whole when inquiring
When number whether there is, the built-in method getbit that Redis database can be used is judged, time complexity is O (1), inquiry speed
Degree is down to constant order by original linear rank.Above-mentioned storage mode not only reduces the memory headroom used, also greatly improves
Query performance, to the intersection of multiple labels, the inquiry of union also can quickly obtain result by bit arithmetic.
S3: user is carried out point group and carries out information according to allocated user group to push away according to user data and label
It send；
It is described that user is carried out point group and carries out information according to allocated user group to push away according to user data and label
It send in step, the tenant group mode carries out a point group in combination with actual operation situation.Such as: it is higher than given threshold for consumption
The push of coupon information is carried out lower than the user of given threshold with consumption, and monitors its consumption.Specifically: definition is daily
The user for consuming M member or more or monthly consuming accumulative N member or more is VIP user；More than the daily consumption m member of definition or monthly
It consumes accumulative n member or more and is channel VIP user for the user of specific channel；The last distance that logs in of definition is now more than 7 days
User be to be lost user, finally log in user in present 4 to 6 days sections be it is pre- be lost user, finally log in distance
User in present 1 to 3 days sections is any active ues etc..
According to allocated user group, implements different coupon information push, customer consumption can be stimulated, raising is stayed
Deposit, such as: when need to test completely subtract certificate, coupons both certificates in marketing to the stimulation degree of customer consumption when, can needle
A/B test is done to same subscriber group, and by test control to test different coupon types to the thorn of customer consumption
Degree is swashed, to formulate optimal discount coupon scheme.For the user group for having pushed coupon information, its consumption is monitored
With retention situation.
Call back message push is carried out for loss user and the user that will be lost, and monitors the step of it recalls situation tool
Body includes: to be lost player according to loss user tag acquisition, will be lost player information, is drawn a portrait according to the user of respective player
Information is formulated call back message and is pushed, wherein it includes but is not limited to short massage notice, coupons that call back message, which pushes mode,
Provide and pay a return visit the mode linked up.Player's progress monitoring operation that middle success wakes up, the operation prison are recalled for user is lost
Control includes statistics and shows the behavior event of the user, including login, search key, supplements with money, consume, for operation personnel
Data visualization, data interaction and data monitoring function are provided.In an other embodiments, which further includes
The variation of monitoring information push front and back number of users is simultaneously shown, and number of users includes: that player seeks advice from number before and after the push
Measure, the number of players that follows up, recall number of players and its recall than etc., convenient effect push to the information platform carries out intuitively
Understand.
Compared with the existing technology, a kind of gaming platform information-pushing method provided by the invention, is effectively combined server
Middle mass users data set, can quickly, accurately position user, it is comprehensive, grasp user characteristics with multi-angle, and by
The user group of distribution carries out information push, realizes the Rapid matching between push content and user preferences, improves matched
Accuracy saves Internet resources；The present invention is analyzed and predicted the loss probability of user by the way of Fusion Model,
The generalization ability of model is greatly improved, customer churn early warning accuracy is high.The present invention also passes through using bitmap algorithm to mark
Label information is stored, and relative to the label storage scheme of traditional structuring, is greatlyd improve in the case where data volume is big
Search efficiency.
The present invention also provides a kind of gaming platform information transmission systems, as shown in figure 5, the gaming platform information pushes
System includes:
Data acquisition module 1, for obtaining the user data in server；
The data acquisition module 1 includes:
Server data obtains module, for obtaining the structuring reported in server database, unstructured data
With report to server local Disk Logs file；
Data processing module, for for unstructured data, the log that user's generation is corresponded in system to be combined,
It is spliced into the text-independent of corresponding user's unique identifier, and is cleaned using participle tool.
Label establishes module 2, for based on user data establish user base information label, customer consumption prediction label,
Game preference label and loss user tag；It includes: that user base information label is established module, used that the label, which establishes module 2,
Family consumption predictions label, which establishes module, game preference label establishes module and is lost user tag establishes module.
Wherein, it includes: that registration behavior label establishes unit that the user base information label, which establishes module, is used for obtaining
Family registion time, registration time length, the information such as registration equipment establish registration behavior label；
User enlivens type label and establishes unit, is used for counting user login time point, calculates user's active period, establishes
User enlivens type label；
City level label where user establishes unit, logs in IP for counting user and parses user's entry address, builds
City level label where vertical user；
User's active index label establishes unit, for calculating user's active index according to user's login time, establishes and uses
Family active index label；
Customer consumption label establishes unit, for participating in movable consumption number of times, spending amount according to user, in conjunction with user
Browsing behavior data establish consumer consumption behavior label and customer consumption index label.
The customer consumption prediction label establishes module and includes:
Data acquisition unit, for being acquired from server log and pre-processing user data；
Characteristic processing unit, for extracting paying customer in a cycle and the user base attributive character without paying customer
With game behavior property feature, and the user base attributive character and game behavior property feature are carried out at Feature Engineering
Reason；
Disaggregated model construction unit, for promoting decision tree building disaggregated model using gradient and rolling over cross validation by K
Method adjusting parameter；
Regression model characteristic processing unit, for extracting the user base attribute and game row of paying customer in a cycle
For attributive character, and Feature Engineering processing is carried out to the user base attribute and game behavior property feature；
Regression model construction unit, for promoting recurrence building regression model using gradient and rolling over cross-validation method by K
Adjusting parameter；
Customer consumption Rulemaking unit, for formulating customer consumption hierarchy rules；
Integral unit, for above-mentioned disaggregated model, regression model and customer consumption hierarchy rules to be integrated into customer consumption
Prediction model.
The game preference label establishes module and includes:
Initial degree of liking computing unit generates corelation behaviour event for counting user before registering each type game
Number, and the initial degree of liking for obtaining corresponding types game is calculated with this；
Growth Rate Calculation unit, time that each type game is put into the period for counting user and upper one
The time of period investment, and calculate with this growth rate of each type playtime increment；
Degree of liking computing unit, for calculating each type of T according to following manneriValue；
Wherein, αiFor the growth rate of user's playtime increment, Δ x is the playtime of user's investment, TjIndicate user
The initial value for degree of liking, ti-tjIndicate time interval, M indicates consumption total value of the user in certain class game, and λ is constant, is indicated
The ratio that spending limit increases temperature, TiIndicate user in tiDegree of liking of the period to certain class game.
User preference determination unit, for being ranked up according to the degree of liking of each type of play in user's period, according to happiness
Love degree value determines the type of play of user preference.
The loss user tag establishes module and includes:
Characteristic data set acquiring unit obtains feature for being associated with user data and label using user's unique identification
Data set；
Characteristic processing unit, for carrying out feature pretreatment to characteristic data set and utilizing gradient boosted tree and feature weight
Carry out feature selecting；
Fusion Model construction unit for training base learning model using K folding cross-validation method, and constructs Fusion Model；
It is lost user tag generation unit, is lost user tag letter for calling Fusion Model identification to be lost user and generate
Breath.
Label memory module 3, for by user base information label, customer consumption prediction label, game preference label and
The bookmark name for being lost user tag is used as key, and the user under label is converted to the corresponding value of the key, by the key and is worth with mutual
Corresponding mode is stored into database；
The label memory module 3 includes:
Acquiring unit, for obtaining unique mark of bookmark name, maximum user identifier and all users with the label
Know；
Converting unit, for the unique identification of users all in label to be converted to k bitmap numbers by bitmap algorithm
Group；Wherein, k=1+N/32, N are user's number of label；
Storage unit, for bitmap array to be converted to hexadecimal string, and using bookmark name as key, described ten
Senary character string is as the corresponding value of the key, and storage is in the database.
Info push module 4, for carrying out point group to user according to user data and label and according to allocated user
Group carries out information push.
The info push module 4 includes:
Push unit is consumed, it is preferential for being carried out higher than given threshold and consumption lower than the user of given threshold for consumption
The push of certificate information, and monitor its consumption；
Unit is recalled in loss, for carrying out call back message push for loss user and the user that will be lost, and is monitored
It recalls number of users；
Monitoring unit pushes the variation of front and back number of users for monitoring information and is shown.
Compared with the existing technology, the present invention provides the modules of whole set of system to cooperate jointly, has built an intelligence
User's operation platform of change, suitable for the platform operation demand of the multiple fields such as electric business, social networks, the gaming platform information
Supplying system has been effectively combined mass users data set, comprehensive, multi-angular analysis user characteristics, by carrying out to user
Divide group and information push carried out according to allocated user group, realizes the Rapid matching between push content and user preferences,
Matched accuracy is improved, Internet resources are saved.
The present invention also provides a kind of computer readable storage mediums, store computer program thereon, the computer program
It realizes when being executed by processor such as the step of above-mentioned gaming platform information-pushing method.
It wherein includes storage medium (the including but not limited to disk of program code that the present invention, which can be used in one or more,
Memory, CD-ROM, optical memory etc.) on the form of computer program product implemented.Computer-readable storage media packet
Permanent and non-permanent, removable and non-removable media is included, can be accomplished by any method or technique information storage.Letter
Breath can be computer readable instructions, data structure, the module of program or other data.The example packet of the storage medium of computer
Include but be not limited to: phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM),
Other kinds of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory
(EEPROM), flash memory or other memory techniques, read-only disc read only memory (CD-ROM) (CD-ROM), digital versatile disc
(DVD) or other optical storage, magnetic cassettes, tape magnetic disk storage or other magnetic storage devices or any other non-biography
Defeated medium, can be used for storage can be accessed by a computing device information.
The present invention also provides a kind of computer equipment, including reservoir, processor and it is stored in the reservoir simultaneously
The computer program that can be executed by the processor, the processor are realized when executing the computer program as above-mentioned game is flat
The step of station information method for pushing.
The invention is not limited to above embodiment, if not departing from the present invention to various changes or deformation of the invention
Spirit and scope, if these changes and deformation belong within the scope of claim and equivalent technologies of the invention, then this hair
It is bright to be also intended to encompass these changes and deformation.
Claims (10)
1. gaming platform information-pushing method, which comprises the following steps:
Obtain the user data in server；
User base information label, customer consumption prediction label, game preference label are established based on user data and are lost user
Label；
Point group is carried out to user according to user data and label and information push is carried out according to allocated user group.
2. gaming platform information-pushing method according to claim 1, it is characterised in that: described to be established based on user data
The step of user base information label includes:
User's registration time, registration time length are obtained, the information such as registration equipment establish registration behavior label；
Counting user logs in IP and parses user's entry address, establishes city level label where user；
Counting user login time point calculates user's active period, establishes user and enliven type label；
According to user's login time, user's active index is calculated, user's active index label is established；
Movable consumption number of times, spending amount are participated according to user, user browsing behavior is analyzed, establishes consumer consumption behavior label
With customer consumption index label.
3. gaming platform information-pushing method according to claim 1, it is characterised in that: described to be established based on user data
The step of customer consumption prediction label includes:
User data is acquired from server log；
Paying customer and user base attributive character and game behavior property feature without paying customer in a cycle are extracted, and
Feature Engineering processing is carried out to the user base attributive character and game behavior property feature；
Decision tree building disaggregated model is promoted using gradient and the parameter that cross-validation method adjusts regression model is rolled over by K；
The user base attribute and game behavior property feature of paying customer in a cycle are extracted, and to the user base category
Property and game behavior property feature carry out Feature Engineering processing；
It is promoted using gradient and returns building regression model and the parameter that cross-validation method adjusts regression model is rolled over by K；
Formulate customer consumption hierarchy rules；
The disaggregated model, regression model and customer consumption hierarchy rules are integrated into customer consumption prediction model.
4. gaming platform information-pushing method according to claim 1, it is characterised in that: described to be established based on user data
The step of game preference label includes:
Counting user generates corelation behaviour event number before registering each type game, and is calculated with this and obtain corresponding types game
Initial degree of liking；
Time that counting user puts into each type game in the period and in the time of investment of the upper period, and in terms of this
Calculate the growth rate of each type playtime；
User's degree of liking of every kind of type of play is calculated according to following manner:
Wherein, αiFor the growth rate of user's playtime increment, Δ x is the playtime of user's investment, TjIndicate that user likes
The initial value of degree, ti-tjIndicate time interval, M indicates consumption total value of the user in certain class game, and λ is constant, indicates consumption
The ratio that amount increases temperature, TiIndicate user in tiDegree of liking of the period to certain class game；
It is ranked up according to degree of liking of the user to type of play each in the period, according to the trip for liking degree value to determine user preference
Play type.
5. gaming platform information-pushing method according to claim 1, it is characterised in that: described to be established based on user data
Be lost user tag the step of include:
User data and label are associated with using user's unique identification, obtain characteristic data set；
Feature selecting is carried out to characteristic data set using gradient boosted tree and feature weight；
If using the trained butt learning model of K folding cross-validation method, and if being melted according to the building of the output result of butt learning model
Molding type；
It calls Fusion Model identification to be lost user and generate and is lost user tag information.
6. gaming platform information-pushing method according to claim 1, it is characterised in that: further comprising the steps of: will use
Family basic information label, customer consumption prediction label, game preference label and the bookmark name for being lost user tag are marked as key
The user signed is converted to the corresponding value of the key, and the key and value are stored in a manner of mutual corresponding into database；Wherein, should
Step specifically includes:
Obtain the unique identification of bookmark name, maximum user identifier and all users with the label；
The unique identification of users all in label is converted to by bitmap algorithm to k bitmap arrays；Wherein, k=1+N/32,
N is user's number of label；
Bitmap array is converted into hexadecimal string, and using bookmark name as key, the hexadecimal string conduct
The corresponding value of the key, storage is in the database.
7. gaming platform information-pushing method according to claim 1, it is characterised in that: described according to user data and mark
Label carry out point group to user and specifically include the step of carrying out information push according to allocated user group:
It is higher than given threshold for consumption and consumes the push for carrying out coupon information lower than the user of given threshold, and monitors it
Consumption；
Call back message push is carried out for loss user and the user that will be lost, and monitors it and recalls number of users；
The variation of monitoring information push front and back number of users is simultaneously shown.
8. a kind of gaming platform information transmission system, it is characterised in that: include:
Data acquisition module, for obtaining the user data in server；
Label establishes module, inclined for establishing user base information label, customer consumption prediction label, game based on user data
Good label and loss user tag；
Info push module, for according to user data and label to user carry out point group and according to allocated user group into
Row information push.
9. a kind of computer readable storage medium, stores computer program thereon, it is characterised in that: the computer program is located
Manage the step of realizing the gaming platform information-pushing method as described in claim 1-7 any one when device executes.
10. a kind of computer equipment, it is characterised in that: including reservoir, processor and be stored in the reservoir and can
The computer program executed by the processor, the processor realize such as claim 1-7 when executing the computer program
Any one of described in gaming platform information-pushing method the step of.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910491711.6A CN110222267B (en) | 2019-06-06 | 2019-06-06 | Game platform information pushing method, system, storage medium and equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910491711.6A CN110222267B (en) | 2019-06-06 | 2019-06-06 | Game platform information pushing method, system, storage medium and equipment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110222267A true CN110222267A (en) | 2019-09-10 |
| CN110222267B CN110222267B (en) | 2023-07-25 |
Family
ID=67815969
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910491711.6A Active CN110222267B (en) | 2019-06-06 | 2019-06-06 | Game platform information pushing method, system, storage medium and equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110222267B (en) |
Cited By (30)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111083211A (en) * | 2019-12-09 | 2020-04-28 | 广州探途网络技术有限公司 | User touch method and system based on big data platform |
| CN111127426A (en) * | 2019-12-23 | 2020-05-08 | 山东大学齐鲁医院 | Gastric mucosa cleanliness evaluation method and system based on deep learning |
| CN111191151A (en) * | 2019-12-20 | 2020-05-22 | 上海淇玥信息技术有限公司 | Method and device for pushing information based on POI (Point of interest) tag and electronic equipment |
| CN111224731A (en) * | 2019-12-26 | 2020-06-02 | 支付宝(杭州)信息技术有限公司 | Content pushing method, device and equipment based on voice broadcast |
| CN111291816A (en) * | 2020-02-17 | 2020-06-16 | 支付宝(杭州)信息技术有限公司 | Method and device for carrying out feature processing aiming at user classification model |
| CN111428129A (en) * | 2020-02-28 | 2020-07-17 | 海南和方信息科技有限公司 | Game information pushing method, system and storage medium thereof |
| CN111737575A (en) * | 2020-06-19 | 2020-10-02 | 北京字节跳动网络技术有限公司 | Content distribution method and device, readable medium and electronic equipment |
| CN111773732A (en) * | 2020-09-04 | 2020-10-16 | 完美世界(北京)软件科技发展有限公司 | Target game user detection method, device and equipment |
| CN111953763A (en) * | 2020-08-06 | 2020-11-17 | 腾讯科技（深圳）有限公司 | Business data pushing method and device and storage medium |
| CN112016961A (en) * | 2020-08-26 | 2020-12-01 | 北京字节跳动网络技术有限公司 | Pushing method and device, electronic equipment and computer readable storage medium |
| CN112015975A (en) * | 2020-07-15 | 2020-12-01 | 北京淇瑀信息科技有限公司 | Financial user-oriented information pushing method and device based on Newton's cooling law |
| CN112426724A (en) * | 2020-11-30 | 2021-03-02 | 北京达佳互联信息技术有限公司 | Game user matching method and device, electronic equipment and storage medium |
| CN112446763A (en) * | 2020-11-27 | 2021-03-05 | 广州三七互娱科技有限公司 | Service recommendation method and device and electronic equipment |
| CN112619165A (en) * | 2020-12-18 | 2021-04-09 | 咪咕互动娱乐有限公司 | Game user selection method, system, server and storage medium |
| CN112651433A (en) * | 2020-12-17 | 2021-04-13 | 广州锦行网络科技有限公司 | Abnormal behavior analysis method for privileged account |
| CN112734463A (en) * | 2020-12-30 | 2021-04-30 | 咪咕音乐有限公司 | Service information sending method and device, electronic equipment and storage medium |
| CN113014476A (en) * | 2021-03-17 | 2021-06-22 | 维沃移动通信有限公司 | Group creation method and device |
| CN113082725A (en) * | 2021-03-08 | 2021-07-09 | 杭州电魂网络科技股份有限公司 | Game user grouping method and device, electronic equipment and storage medium |
| CN113329058A (en) * | 2021-04-30 | 2021-08-31 | 青岛以萨数据技术有限公司 | Data pushing method and device and storage medium |
| CN113413610A (en) * | 2021-06-21 | 2021-09-21 | 网易（杭州）网络有限公司 | Game item recommendation method, equipment and computer-readable storage medium |
| CN113572753A (en) * | 2021-07-16 | 2021-10-29 | 北京淇瑀信息科技有限公司 | User equipment authentication method and device based on Newton's cooling law |
| CN113706202A (en) * | 2021-08-31 | 2021-11-26 | 杭州群核信息技术有限公司 | Recall strategy generating method based on low-steady-state user identification and early warning |
| CN113742388A (en) * | 2020-05-29 | 2021-12-03 | 北京顺源开华科技有限公司 | Data pushing method and device, storage medium and electronic equipment |
| CN113837780A (en) * | 2020-06-23 | 2021-12-24 | 上海莉莉丝科技股份有限公司 | Information delivery method, system, device and medium |
| CN114832386A (en) * | 2022-04-26 | 2022-08-02 | 江苏果米文化发展有限公司 | Game user intelligent management system based on big data analysis |
| TWI776742B (en) * | 2021-11-29 | 2022-09-01 | 愛酷智能科技股份有限公司 | System for analyzing user behavior in information exchange platform |
| CN115168740A (en) * | 2022-09-06 | 2022-10-11 | 网娱互动科技(北京)股份有限公司 | Method and system for generating marketing task based on big data analysis |
| CN115531886A (en) * | 2022-10-08 | 2022-12-30 | 广州易幻网络科技有限公司 | User and equipment data management method, system and storage medium |
| CN116492693A (en) * | 2023-06-28 | 2023-07-28 | 北京乐盟互动科技有限公司 | Method and device for promoting online game data, electronic equipment and storage medium |
| CN116757750A (en) * | 2023-06-05 | 2023-09-15 | 广州盈风网络科技有限公司 | Operation pushing method, device, equipment and medium based on loss rate prediction |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105912729A (en) * | 2016-05-19 | 2016-08-31 | 华南理工大学 | Three-dimensional time sequence dynamic model based mobile phone game recommendation system |
| CN106097114A (en) * | 2016-06-27 | 2016-11-09 | 江苏易乐网络科技有限公司 | A kind of game public sentiment monitoring and player's behavior analysis cloud service platform |
| CN107203899A (en) * | 2017-05-03 | 2017-09-26 | 北京小米移动软件有限公司 | Information-pushing method and equipment |
| CN107609708A (en) * | 2017-09-25 | 2018-01-19 | 广州赫炎大数据科技有限公司 | A kind of customer loss Forecasting Methodology and system based on mobile phone games shop |
-
2019
- 2019-06-06 CN CN201910491711.6A patent/CN110222267B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105912729A (en) * | 2016-05-19 | 2016-08-31 | 华南理工大学 | Three-dimensional time sequence dynamic model based mobile phone game recommendation system |
| CN106097114A (en) * | 2016-06-27 | 2016-11-09 | 江苏易乐网络科技有限公司 | A kind of game public sentiment monitoring and player's behavior analysis cloud service platform |
| CN107203899A (en) * | 2017-05-03 | 2017-09-26 | 北京小米移动软件有限公司 | Information-pushing method and equipment |
| CN107609708A (en) * | 2017-09-25 | 2018-01-19 | 广州赫炎大数据科技有限公司 | A kind of customer loss Forecasting Methodology and system based on mobile phone games shop |
Cited By (44)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111083211A (en) * | 2019-12-09 | 2020-04-28 | 广州探途网络技术有限公司 | User touch method and system based on big data platform |
| CN111083211B (en) * | 2019-12-09 | 2023-08-11 | 广州探途网络技术有限公司 | User touch method and system based on big data platform |
| CN111191151A (en) * | 2019-12-20 | 2020-05-22 | 上海淇玥信息技术有限公司 | Method and device for pushing information based on POI (Point of interest) tag and electronic equipment |
| CN111191151B (en) * | 2019-12-20 | 2023-08-25 | 上海淇玥信息技术有限公司 | Method and device for pushing information based on POI (point of interest) tag and electronic equipment |
| CN111127426A (en) * | 2019-12-23 | 2020-05-08 | 山东大学齐鲁医院 | Gastric mucosa cleanliness evaluation method and system based on deep learning |
| CN111127426B (en) * | 2019-12-23 | 2020-12-01 | 山东大学齐鲁医院 | Gastric mucosa cleanliness evaluation method and system based on deep learning |
| CN111224731A (en) * | 2019-12-26 | 2020-06-02 | 支付宝(杭州)信息技术有限公司 | Content pushing method, device and equipment based on voice broadcast |
| CN111224731B (en) * | 2019-12-26 | 2021-06-08 | 支付宝(杭州)信息技术有限公司 | Content pushing method, device and equipment based on voice broadcast |
| CN111291816A (en) * | 2020-02-17 | 2020-06-16 | 支付宝(杭州)信息技术有限公司 | Method and device for carrying out feature processing aiming at user classification model |
| CN111291816B (en) * | 2020-02-17 | 2021-08-06 | 支付宝(杭州)信息技术有限公司 | Method and device for carrying out feature processing aiming at user classification model |
| CN111428129A (en) * | 2020-02-28 | 2020-07-17 | 海南和方信息科技有限公司 | Game information pushing method, system and storage medium thereof |
| CN113742388A (en) * | 2020-05-29 | 2021-12-03 | 北京顺源开华科技有限公司 | Data pushing method and device, storage medium and electronic equipment |
| CN111737575B (en) * | 2020-06-19 | 2023-11-14 | 北京字节跳动网络技术有限公司 | Content distribution method, content distribution device, readable medium and electronic equipment |
| CN111737575A (en) * | 2020-06-19 | 2020-10-02 | 北京字节跳动网络技术有限公司 | Content distribution method and device, readable medium and electronic equipment |
| CN113837780A (en) * | 2020-06-23 | 2021-12-24 | 上海莉莉丝科技股份有限公司 | Information delivery method, system, device and medium |
| CN112015975A (en) * | 2020-07-15 | 2020-12-01 | 北京淇瑀信息科技有限公司 | Financial user-oriented information pushing method and device based on Newton's cooling law |
| CN112015975B (en) * | 2020-07-15 | 2023-11-14 | 北京淇瑀信息科技有限公司 | Information pushing method and device for financial users based on Newton's law of cooling |
| CN111953763A (en) * | 2020-08-06 | 2020-11-17 | 腾讯科技（深圳）有限公司 | Business data pushing method and device and storage medium |
| CN111953763B (en) * | 2020-08-06 | 2021-06-25 | 腾讯科技（深圳）有限公司 | Business data pushing method and device and storage medium |
| CN112016961A (en) * | 2020-08-26 | 2020-12-01 | 北京字节跳动网络技术有限公司 | Pushing method and device, electronic equipment and computer readable storage medium |
| CN111773732A (en) * | 2020-09-04 | 2020-10-16 | 完美世界(北京)软件科技发展有限公司 | Target game user detection method, device and equipment |
| CN112446763A (en) * | 2020-11-27 | 2021-03-05 | 广州三七互娱科技有限公司 | Service recommendation method and device and electronic equipment |
| CN112426724A (en) * | 2020-11-30 | 2021-03-02 | 北京达佳互联信息技术有限公司 | Game user matching method and device, electronic equipment and storage medium |
| CN112426724B (en) * | 2020-11-30 | 2023-07-25 | 北京达佳互联信息技术有限公司 | Matching method and device for game users, electronic equipment and storage medium |
| CN112651433A (en) * | 2020-12-17 | 2021-04-13 | 广州锦行网络科技有限公司 | Abnormal behavior analysis method for privileged account |
| CN112651433B (en) * | 2020-12-17 | 2021-12-14 | 广州锦行网络科技有限公司 | Abnormal behavior analysis method for privileged account |
| CN112619165A (en) * | 2020-12-18 | 2021-04-09 | 咪咕互动娱乐有限公司 | Game user selection method, system, server and storage medium |
| CN112734463A (en) * | 2020-12-30 | 2021-04-30 | 咪咕音乐有限公司 | Service information sending method and device, electronic equipment and storage medium |
| CN113082725A (en) * | 2021-03-08 | 2021-07-09 | 杭州电魂网络科技股份有限公司 | Game user grouping method and device, electronic equipment and storage medium |
| CN113014476A (en) * | 2021-03-17 | 2021-06-22 | 维沃移动通信有限公司 | Group creation method and device |
| CN113014476B (en) * | 2021-03-17 | 2023-04-07 | 维沃移动通信有限公司 | Group creation method and device |
| CN113329058A (en) * | 2021-04-30 | 2021-08-31 | 青岛以萨数据技术有限公司 | Data pushing method and device and storage medium |
| CN113329058B (en) * | 2021-04-30 | 2022-10-04 | 青岛以萨数据技术有限公司 | Data pushing method and device and storage medium |
| CN113413610A (en) * | 2021-06-21 | 2021-09-21 | 网易（杭州）网络有限公司 | Game item recommendation method, equipment and computer-readable storage medium |
| CN113572753B (en) * | 2021-07-16 | 2023-03-14 | 北京淇瑀信息科技有限公司 | User equipment authentication method and device based on Newton's cooling law |
| CN113572753A (en) * | 2021-07-16 | 2021-10-29 | 北京淇瑀信息科技有限公司 | User equipment authentication method and device based on Newton's cooling law |
| CN113706202A (en) * | 2021-08-31 | 2021-11-26 | 杭州群核信息技术有限公司 | Recall strategy generating method based on low-steady-state user identification and early warning |
| TWI776742B (en) * | 2021-11-29 | 2022-09-01 | 愛酷智能科技股份有限公司 | System for analyzing user behavior in information exchange platform |
| CN114832386A (en) * | 2022-04-26 | 2022-08-02 | 江苏果米文化发展有限公司 | Game user intelligent management system based on big data analysis |
| CN114832386B (en) * | 2022-04-26 | 2024-05-14 | 江苏果米文化发展有限公司 | Game user intelligent management system based on big data analysis |
| CN115168740A (en) * | 2022-09-06 | 2022-10-11 | 网娱互动科技(北京)股份有限公司 | Method and system for generating marketing task based on big data analysis |
| CN115531886A (en) * | 2022-10-08 | 2022-12-30 | 广州易幻网络科技有限公司 | User and equipment data management method, system and storage medium |
| CN116757750A (en) * | 2023-06-05 | 2023-09-15 | 广州盈风网络科技有限公司 | Operation pushing method, device, equipment and medium based on loss rate prediction |
| CN116492693A (en) * | 2023-06-28 | 2023-07-28 | 北京乐盟互动科技有限公司 | Method and device for promoting online game data, electronic equipment and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110222267B (en) | 2023-07-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110222267A (en) | A kind of gaming platform information-pushing method, system, storage medium and equipment | |
| Neslin et al. | Defection detection: Measuring and understanding the predictive accuracy of customer churn models | |
| CN103714139B (en) | Parallel data mining method for identifying a mass of mobile client bases | |
| CN103533390B (en) | The method and system of television program recommendations are carried out based on social network information | |
| CN104834686A (en) | Video recommendation method based on hybrid semantic matrix | |
| CN103177129B (en) | Internet real-time information recommendation prognoses system | |
| CN112598438A (en) | Outdoor advertisement recommendation system and method based on large-scale user portrait | |
| CN106951471A (en) | A kind of construction method of the label prediction of the development trend model based on SVM | |
| Gan | XGBoost-based e-commerce customer loss prediction | |
| CN104572915A (en) | User event relevance calculation method based on content environment enhancement | |
| TWI657395B (en) | Opinion leader related network-based trading system, method, and storage medium | |
| KR102238438B1 (en) | System for providing commercial product transaction service using price standardization | |
| Lei et al. | Research on the evaluation of cross-border e-commerce overseas strategic climate based on decision tree and adaptive boosting classification models | |
| CN113052653A (en) | Financial product content recommendation method and system and computer readable storage medium | |
| Wang et al. | Research on financial advertisement personalised recommendation method based on customer segmentation | |
| Li et al. | A personalized recommendation algorithm for college books based on user interest | |
| Zhang et al. | A recommender system for cold-start items: a case study in the real estate industry | |
| Yan et al. | An E-commerce coupon target population positioning model based on random forest and extreme gradient boosting | |
| Huang et al. | Spanning 36 years, the evolution and trend of word-of-mouth marketing research–based on bibliometrix analysis | |
| Chen et al. | Dynamic box office forecasting based on microblog data | |
| Zu | The validity of visual communication design of online advertisement based on interactivity | |
| CN113362034A (en) | Position recommendation method | |
| Nune | Characterizing and predicting early reviewers for effective product marketing on ecommerce websites | |
| Krykavskyy et al. | The features of anticipative marketing in the process of products promotion. | |
| Li et al. | A Method of Interest Degree Mining Based on Behavior Data Analysis |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |