CN110222267B - Game platform information pushing method, system, storage medium and equipment - Google Patents
Game platform information pushing method, system, storage medium and equipment Download PDFInfo
- Publication number
- CN110222267B CN110222267B CN201910491711.6A CN201910491711A CN110222267B CN 110222267 B CN110222267 B CN 110222267B CN 201910491711 A CN201910491711 A CN 201910491711A CN 110222267 B CN110222267 B CN 110222267B
- Authority
- CN
- China
- Prior art keywords
- user
- game
- label
- users
- 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.)
- Active
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 invention relates to a game platform information pushing method, a game platform information pushing system, a game platform information pushing storage medium and game platform information pushing equipment. Compared with the prior art, the method and the device realize quick matching between the push content and the user demand by grouping the users and pushing the information according to the allocated user group, improve the matching accuracy and save the network resources.
Description
Technical Field
The present invention relates to the field of game operation, and in particular, to a method, a system, a storage medium, and a device for pushing game platform information.
Background
With the rapid development of internet technology, the game industry is also gradually changed, various game layers are endless, competition among game operators is more and more vigorous, and under the condition that the game experience is not greatly different, stable operation and good service determine whether a game can stand out. The cost of acquiring new users is higher and higher, and the loss of users is a common phenomenon in game operation, so that in order to maintain the number of users, the game platform can push game information to the users to ensure the viscosity of the users.
However, in the actual service, when the game platform pushes information to users, information is often pushed to all users, so that the situation that the pushed information does not accord with the actual demand of the users easily occurs, and the excessive irrelevant pushed information occupies network resources and lacks attractive force to the users due to mismatching with the demand of the users, so that the effect of the pushed information is poor.
Disclosure of Invention
The invention aims to overcome the defects and shortcomings of the prior art and provide a game platform information pushing method, a game platform information pushing system, a game platform information storage medium and game platform information pushing equipment, wherein pushing content is matched with user requirements, and the game platform information pushing method, the game platform information pushing system, the game platform information storage medium and the game platform information storage device are high in matching performance.
A game platform information pushing method comprises the following steps:
acquiring user data in a server;
establishing a user basic information label, a user consumption prediction label, a game preference label and a loss user label based on the user data;
grouping users according to the user data and the labels and pushing information according to the allocated user groups.
Compared with the prior art, the scheme can help operators to quickly and accurately position users by analyzing data of a large amount of user data in the server and establishing corresponding labels, master user characteristics in all directions and multiple angles, and improve matching accuracy and save network resources by grouping the users and pushing information according to allocated user groups.
Further, the step of creating a user base information tag based on the user data includes:
acquiring user registration time, registration duration, registration equipment and other information to establish a registration behavior label;
counting user login IP, resolving user login address, and establishing city grade label of user;
counting user login time points, calculating user activity time periods, and establishing user activity type labels;
calculating a user activity index according to the user login time, and establishing a user activity index label;
and analyzing the browsing behaviors of the user according to the consumption times and the consumption amount of the user participating in the activities, and establishing a user consumption behavior label and a user consumption index label. By establishing the basic information label of the user, the user can be conveniently analyzed in an omnibearing and multi-angle manner, and the quick matching between the push content and the preference of the user is realized.
Further, the step of establishing a user consumption prediction tag based on the user data includes:
collecting user data from a server log;
extracting user basic attribute characteristics and game behavior attribute characteristics of paid users and unpaid users in a period, and carrying out feature engineering processing on the user basic attribute characteristics and the game behavior attribute characteristics;
Constructing a classification model by utilizing a gradient lifting decision tree and adjusting parameters of the classification model by a K-fold cross validation method;
extracting user basic attribute characteristics and game behavior attribute characteristics of a paying user in a period, and carrying out feature engineering processing on the user basic attribute characteristics and the game behavior attribute characteristics;
constructing a regression model by utilizing gradient lifting regression and adjusting parameters of the regression model by a K-fold cross validation method;
formulating a user consumption level rule;
and integrating the classification model, the regression model and the user consumption level rule into a user consumption prediction model. The regression model is constructed by using gradient lifting regression, so that various types of data, including discrete characteristics and continuous characteristics, can be flexibly processed, and the robustness to abnormal values is higher and the accuracy is higher.
Further, the step of establishing a game preference tag based on the user data includes:
counting the number of related behavior events generated by a user before registering each type of game, and calculating to obtain the initial favorites of the corresponding type of game;
counting the time input by the user to each type of game in the period and the time input in the previous period, and calculating the growth rate of the growth amount of each type of game time according to the time input by the user in the previous period;
The user preference for each game type is calculated according to the following manner:
wherein alpha is i For the increase rate of the increase amount of the game time of the user, deltax is the game time input by the user, T j Initial value, t, representing user preference i -t j Represents time interval, M represents total consumption of users in a certain game, lambda is constant, represents proportion of consumption to heat increase, T i Representing the user at t i The period favorites for a certain class of games. By constructing the user game preference labels, subsequent operators can conveniently implement specific marketing strategies for specified groups of people.
Further, the step of establishing the churn user tag based on the user data includes:
associating the user data with the tag by using the unique user identifier to obtain a characteristic data set;
performing feature selection on the feature data set by utilizing the gradient lifting tree and the feature weight;
training a plurality of basic learning models by using a K-fold cross validation method, and constructing a fusion model according to the output results of the plurality of basic learning models;
and calling the fusion model to identify the loss user and generating the label information of the loss user. The fusion model is called to identify the lost user by learning various behavior characteristics of different users by utilizing various single models, so that the prediction accuracy of the lost user is improved.
Further, the method also comprises the following steps: the label names of the user basic information labels, the user consumption prediction labels, the game preference labels and the loss user labels are used as keys, the users under the labels are converted into values corresponding to the keys, and the keys and the values are stored in a database in a mutually corresponding mode; wherein, this step specifically includes:
acquiring a tag name, a maximum user identifier and unique identifiers of all users with the tag;
converting unique identifications of all users in the tag into a k-bit bitmap array through a bitmap algorithm; wherein k=1+n/32, n being the number of users of the tag;
the bitmap array is converted into hexadecimal character strings, the label name is used as a key, and the hexadecimal character strings are used as the corresponding values of the key and are stored in a database. The bitmap algorithm is utilized to store the label information, so that the used memory space is reduced, and the query performance is greatly improved.
Further, the step of grouping the users according to the user data and the labels and pushing information according to the allocated user groups specifically includes:
pushing coupon information aiming at users with consumption higher than a set threshold value and consumption lower than the set threshold value, and monitoring the consumption condition of the users;
Carrying out recall information pushing aiming at lost users and users to be lost, and monitoring the number of recall users;
and monitoring and displaying the change of the number of users before and after information pushing. By pushing different information and monitoring the number change of users, the information pushing effect of the game platform is intuitively displayed, and data support is provided for subsequent activities.
The invention also provides a game platform information pushing system, which comprises:
the data acquisition module is used for acquiring user data in the server;
the label building module is used for building a user basic information label, a user consumption prediction label, a game preference label and an attrition user label based on user data;
and the information pushing module is used for grouping the users according to the user data and the labels and pushing the information according to the allocated user groups.
Compared with the prior art, the scheme provides a whole set of systematic module co-operation, and by carrying out data analysis on a large amount of user data in the server and establishing corresponding labels, operators can be helped to quickly and accurately position users, master the characteristics of the users in all directions and at multiple angles, realize quick matching between push content and user preference, improve matching accuracy and save network resources.
The present invention also provides a computer-readable storage medium having stored thereon a computer program characterized in that: the computer program when executed by the processor implements the steps of the game platform information pushing method described above.
The invention also provides a computer device, which comprises a storage, a processor and a computer program stored in the storage and executable by the processor, wherein the steps of the game platform information pushing method are realized when the processor executes the computer program.
For a better understanding and implementation, the present invention is described in detail below with reference to the drawings.
Drawings
FIG. 1 is a flowchart of a game platform information pushing method according to embodiment 1 of the present invention;
FIG. 2 is a schematic diagram of a game platform information pushing method according to embodiment 1 of the present invention;
FIG. 3 is a flow chart of the construction of a fusion model using K-fold cross-validation in embodiment 1 of the present invention;
FIG. 4 is a block diagram of a fusion model in embodiment 1 of the present invention;
fig. 5 is a block diagram of a game platform information push system according to embodiment 2 of the present invention.
Detailed Description
Example 1
Fig. 1-2 are flowcharts of a game platform information pushing method according to the present invention.
A game platform information pushing method comprises the following steps:
s1: acquiring user data in a server;
in the step of obtaining the user data in the server, the user data in the server refers to data (including but not limited to platform product data and third party data collected by a crawler) reported to the server by multiple products, and the user overall data set obtained by an ETL (Extract-Transform-Load) technology specifically includes:
reporting structured and unstructured data in a server database and reporting the structured and unstructured data to a server local disk log file; the database includes, but is not limited to, mySQL, mongoDB, HBase database, wherein the structured and unstructured data reported to the server database includes a user registry, a user login behavior table, a user order table, a user recharging table, a user browsing list and the like, and specifically includes structured data such as a user unique identifier, a user age, a registration time, a login time, a consumption amount, a coupon consumption list and the like; and reporting the unstructured data such as the log file generated when the user is followed to the server local disk log file. And for unstructured data, the logs generated by the corresponding users in the system are combined and spliced into an independent text corresponding to the unique identifiers of the users, and the independent text is cleaned by using a word segmentation tool. Specifically, word segmentation is performed by using a word segmentation tool, stop words are removed, a corpus is obtained, then a user keyword is obtained through a TF-IDF algorithm, the TF-IDF algorithm is commonly used for obtaining the importance degree of words in a file, and the calculation process is not particularly limited in the embodiment. The word segmentation tools in this step include, but are not limited to, a segmentation tool such as CoreNLP, jieba.
S2: user base information tags, user consumption prediction tags, game preference tags, and churn user tags are established based on the user data. In other embodiments, the user can increase or reestablish other user tag information according to the actual requirement in this step, so as to facilitate subsequent exploration of more user tag information and maximize the data-driven value.
Wherein, data analysis is performed on the user data, and the step of establishing the user basic information label comprises the following steps:
s211: acquiring user registration time, registration duration, registration equipment and other information to establish a registration behavior label;
s212: counting user login IP, resolving user login address, and establishing city grade label of user;
s213: counting user login time points, calculating user activity time periods, and establishing user activity type labels; wherein the active type comprises a daytime active type or a nighttime active type.
S214: according to the user login time, calculating a user activity index, and establishing a user activity index label, wherein the user activity index can be comprehensively evaluated according to time granularity, such as the login days of near T days, the offline days of near T days and the like, and specifically, the user activity index is obtained by the following steps:
Wherein, activeIndex is user activity index, w i The weight value of the active index is T, the active period of the user is X i The method comprises the steps that an index is active for a user in a period T, and N is the number of events for evaluating the active index; for example, in the present embodiment, n=3, x 1 For offline days in period, X 2 For login days in period, X 3 Is the difference between the last login date and the first login date in the period.
The user liveness index measurement labels are:
the activeindexlbel is an liveness index, when the liveness index is 1, the user is low liveness, when the liveness index is 2, the user is generally lively, and when the liveness index is 3, the user is highly lively.
S215: and according to the consumption times and the consumption amount of the user participating in the activity, combining the user browsing behavior data, and establishing a user consumption behavior label and a user consumption index label.
Wherein, establishing the user consumption behavior label in step S215 includes:
establishing a marketing activity sensitivity index label of the user by combining the ratio of the consumption times of the user participating in the activity to the total consumption times;
and (3) carrying out qualitative analysis on the rational consumption index of the user in combination with the user browsing behavior data, for example, analyzing the relation between the number of times the user clicks a detail page entering the purchasing equipment, the number of browsing equipment and the time period from the first browsing to the purchasing equipment, and qualifying the user as impulse consumption type, rational consumption type, hesitation consumption type and the like, for example, when the user firstly enters the detail page of the equipment and does not browse other detail pages of the equipment, purchasing the equipment, and qualifying the equipment as impulse consumption type and the like.
In step S215, the step of creating a user consumption index tag includes:
counting the consumption frequency, the latest consumption time, the total consumption amount, the maximum consumption amount, the special price consumption frequency and the high price consumption frequency of all users as a data set;
calculating quartiles of each type of consumption attribute in the data set, and grading the consumption attributes of the users according to the size relation between the attribute values and the quartiles; the scoring items include a frequency of consumption score, a last time of consumption score, a total amount of consumption score, a maximum amount of consumption score, a preferential frequency of consumption score, and a high-priced frequency of consumption score.
The user consumption index is obtained by weighting and calculating each attribute scoring item,
wherein w is i Respectively scoring weights of each item, i is the number of scoring items, i is [1,2,3,4,5,6 ]]。
And establishing a user consumption index label according to the user consumption index.
Specifically, in step S2, the step of performing data analysis on the user data, and establishing the user consumption prediction tag includes:
s221: collecting and preprocessing user data from a server log; wherein the preprocessing user data step comprises:
associating a user device number with a user account, filtering abnormal users associated with the device and filtering device numbers associated with a plurality of user accounts, wherein such users may be abnormal users such as generation exercises and the like.
Filtering the abnormal payment users;
ignoring attributes for which the attribute value is largely missing;
performing independent heat coding on discrete features such as channel types, mobile phone platforms and the like;
quantifying user game events;
binarizing the basic attributes of the users such as mobile phone numbers, identity cards and the like.
S222: extracting user basic attribute characteristics and game behavior attribute characteristics of paid users and unpaid users in a period, and carrying out feature engineering processing on the user basic attribute characteristics and the game behavior attribute characteristics;
wherein the user base attributes include: gender, age, platform type, registration channel, registered days, login days, associated account number, VIP grade, login times and common IP area; the user game behavior attribute includes: the number of user games, the type of user games, the user game rating, the user game role rating, the user game rating, the number of game events, the number of important activities involved, the duration of the game; the step of carrying out feature engineering processing on the user basic attribute features and the game behavior attribute features specifically comprises the following steps:
classifying game events and counting the total number of each type of event; filtering the features with large linear relation when PCA or other dimension reduction methods are applied to processing, and simultaneously calculating the mean value, the median value and the like of the similar features as new features to be added into a data set; filtering attributes with variance of 0 or close to 0;
And scoring the features by using the RF value, the F value and the mutual information value, and selecting the intersection of the ten user features before three scores as the key features of the user classification model.
S223: constructing a classification model by utilizing a gradient lifting decision tree and adjusting parameters by a K-fold cross validation method; specifically, a classification model is built by using a gradient lifting decision tree (Gradient Boosting Decision Tree, GBDT), and an optimal classification model is obtained through a K-fold cross validation method and is used for predicting whether a user pays or not.
S224: extracting user basic attributes and game behavior attribute characteristics of a paying user in a period, and carrying out feature engineering processing on the user basic attributes and the game behavior attribute characteristics; specifically, the user basic attribute, the user consumption behavior attribute and the user game behavior attribute of the paying user in the period T in the user data set are extracted, and the characteristic engineering processing is carried out on the user basic attribute characteristic and the game behavior attribute characteristic by using the method described in S222. Wherein the user consumption behavior attributes include: the total amount of consumption, the number of consumption, the last consumption time in a period, the last consumption amount in a period, the average consumption amount, the high-unit price commodity consumption frequency and the maximum consumption amount.
S225: constructing a regression model by utilizing gradient lifting regression and adjusting parameters by a K-fold cross validation method; in a preferred embodiment, gradient lifting regression (Gradient boosting regression, GBR) is used to build a regression model, and the optimal regression model is obtained by a K-fold cross validation method to predict the payment amount of the user.
K-1 is selected as training sets each time, and the rest 1 is used as test sets, so that K groups of training sets and test sets are obtained. Selecting a group of training sets, training a base learner by using the initial weights, and updating weights in samples according to the error rate of the base learner, so that the weights of training sample points with high error rate in the base learner become high, namely, the samples with high error rate are valued in the next base learner; then training the next basic learner by using the training set after the weight adjustment, and cycling until the number of the basic learners reaches the designated number T; finally, combining the T basic learners to obtain a strong learner; and traversing the K groups of training sets to obtain K groups of strong learners, and selecting the strong learner with the minimum mean square error from the K groups of strong learners. By using the gradient lifting strategy, various types of data, including discrete features and continuous features, can be flexibly processed, the robustness to abnormal values is high, and meanwhile, the accuracy is higher than that obtained by using a single model.
S226: formulating a user consumption level rule; specifically, the score of the paid user group in the user data set period T is calculated, 5 user consumption intervals are formulated according to the score, the user consumption level of each interval is defined, and the user consumption levels are sequentially defined as 1 to 5 levels in the order from small to large.
S227: and integrating the classification model, the regression model and the user consumption level rule into a user consumption prediction model. The user consumption prediction model can predict whether the user pays in the future N days, and for the user with the predicted result of payment, the predicted payment amount is obtained through the regression model, and the payment amount is mapped into the user consumption grade and output.
In the step of creating the game preference label based on the user data, since the preference of the user to a certain type of game is generally time-efficient, that is, the preference degree of the user decreases with the increase of time, the present invention adopts an improved time decay function model based on Newton's law of cooling, and calculates the game preference label of the user based on the time decay function model, and specifically, the present invention includes the following steps:
s231: counting the number of relevant behavior events generated by the user before registering each type of game, wherein the relevant behavior comprises: searching games, clicking game introduction, forums, evaluations and the like in the platform; every time a related event record is generated, the corresponding game type is initially favored by T j Increasing a unit value;
s232: counting the time input by the user to each type of game in the period and the time input in the previous period, and calculating the growth rate of the growth amount of each type of game time according to the time input by the user in the previous period; if the user recently likes to play a certain type of game, a great deal of time is invested, and the operation is frequent and the behavior is rich; if the user loses interest in the game, the game time and operation behavior are gradually reduced. Thus, the increase rate α of the increase amount of the user game time i Can be expressed as:
wherein t is j For a certain period t i For another game cycle, alpha i For the rate of increase, deltaX, of the amount of increase in the user's game time j For the user in period t j Game time of betting, Δx i For the user in period t i Game time of the betting.
If alpha is i If the result is positive, the preference of the user is increased; and inversely decreases.
S233: the user preference T for each game type is calculated according to the following manner i Value:
wherein alpha is i For the increase rate of the increase amount of the game time of the user, deltax is the game time input by the user, T j Initial value of user preference, t i -t j Represents time interval, M represents total consumption of users in a certain game, lambda is constant, represents proportion of consumption to heat increase, T i Representing the user at t i The period favorites for a certain class of games.
S234: and ordering the favorites of the game types in the period according to the user, and determining the game types preferred by the user according to the favorites value. The calculation rule of the initial favorites of the user can be formulated according to specific application scenes; for the same batch of game types with small difference of result values calculated by the initial preference rule, the subsequent period can be used for ranking the preference degree of the user on the same batch of game types by directly comparing the T values.
The step of creating a churn user tag based on user data includes:
s241: associating the user data with the tag by using the unique user identifier to obtain a characteristic data set; specifically, the user data set and the tag data are integrated, and the feature data set is obtained through the association of the unique user identification. The user data set comprises fields including but not limited to user registration age, registration IP, login times of near N days, comment times of near N days, praise times of near N days and the like; based on the tag creation module, the tag data includes, but is not limited to, fields such as city level of user login, user consumption type, consumption index, game preference value, etc.
S242: performing feature selection on the feature data set by utilizing the gradient lifting tree and the feature weight;
in one embodiment, before the step of performing feature selection on the feature data set by using the gradient lifting tree and the feature weight, the step of performing feature preprocessing on the feature data set includes: discarding or filling the default value, and filling the qualitative feature according to 0 and the quantitative feature value according to the average value of the qualitative feature value for discarding samples with large default value ratio; for different dimensions, the same specification is ensured according to non-dimensionalization treatment, and the comparison is convenient. The dimensionless processing mode of data common in the prior art such as standardization, interval scaling and the like can be adopted for the dimensionless processing mode; for information redundancy, dividing effective information contained in the information according to intervals, such as near-N-day online time length, only concerning whether the online time length reaches a certain threshold value, and processing the information into 0 and 1 for representing the threshold value which is not reached and the threshold value which is reached;
in the step of performing feature selection by using a gradient lifting tree and feature weights, a feature is selected according to the feature weights under the condition of ensuring the optimal accuracy by using a tree-based feature selection algorithm, wherein the feature is selected according to the feature weights by examining the correlation between the feature and a target value, and the feature is trained by adopting a machine learning method to obtain the weight coefficient of each feature and the feature with lower shift-out weight coefficient. In other embodiments, this step may also utilize a filtering method (Filter) to perform feature selection by examining whether the features diverge and scoring each feature according to its divergence, setting a threshold or the number of thresholds to be selected.
S243: training a plurality of basic learning models by using a K-fold cross validation method, and constructing a fusion model according to the output results of the plurality of basic learning models; specifically, a Stacking-based fusion method is adopted to improve the prediction capability of the model, and compared with a single prediction model, the fusion model has the advantages that the fusion model can combine various behavior characteristics of different users learned by various single models, and can also show good robustness in various environments, and the method for constructing the fusion model in the step comprises the following steps:
the data set after feature selection and feature pretreatment is divided into N equal parts, wherein each basic learning model is obtained by training N-1 parts of data sets, the rest 1 parts are used as test sets, and the prediction results of all the basic learning models are used as training sets and are used as input of the next step. Among the base learning models include, but are not limited to, XGBoost, random Forest (RF), logistic regression (Logistic Regression, LR), and Neural Networks (NN). In a preferred embodiment, a K-fold cross-validation method is used to distinguish between test sets and training sets of different models when training a base learning model, so as to increase the difference between models and improve the fusion effect of the models, and the method specifically comprises the following steps: traversing the N-1 training sets, dividing each traversed training set into K equal parts, selecting K-1 training sets each time, and taking the rest 1 training sets as test sets to obtain K groups of training sets and test sets; k different basic learning models are respectively adopted for training the K groups of training sets and the test sets, such as XGBoost, RF, LR and the like, as shown in figure 3; and taking the prediction results of the K base learning models trained out of the models as the input of the fusion of the second-layer models. The K different base learning models can be combined with actual prediction results, and models with poor partial effects are abandoned.
S244: and calling the fusion model to identify the loss user and generating the label information of the loss user.
And taking the user loss probability value output by the base learning model as the input of the fusion model. As shown in fig. 4, XGBoost and LR are respectively used in the fusion layer to train and predict the user loss probability, and in one embodiment, the probability values output by the two models are averaged to be used as the final output probability of the fusion layer. In other embodiments, the probability values output by the two models are weighted average as the final output probability of the fusion layer. And when the probability average value is larger than the set threshold value, judging the user as the lost user.
And inputting the data set to be predicted, calling the fusion model to calculate, and outputting a user loss list as loss user label information.
In one embodiment, the game platform information pushing method further includes the following steps: taking the label name as a key, and storing the user basic information label, the user consumption prediction label, the game preference label and the loss user label into a database in a mode of converting the user under the label into a value corresponding to the key; in this step, label information is stored in a database by using a bitmap algorithm, that is, a Value (Value) corresponding to an element is marked by using a bit, and a Key (Key) is used as the element itself, which includes the following steps:
Label name lable_k, maximum user identification M and all users with the label obtained according to the label model:
<uid_1,uid_2,…,uid_n>；
wherein, all user unique identifiers are mapped into integer values of self-increasing sequences, and uid_1, uid_2 and …, uid_n are user unique identifiers of all users under the tag;
converting < uid_1, uid_2, …, uid_n > into a bitmap array L by a bitmap algorithm, wherein L is:
<b_uid_1,b_uid_2,…,b_uid_k>；
where k=1+n/32, and k is the number of bitmap arrays.
The bitmap array L is converted into a hexadecimal character string, and the label name lable_k is used as a Key (Key), and the hexadecimal character string is stored in a database as a Value (Value) corresponding to the Key. Wherein the database includes, but is not limited to, a Redis, mySQL, mongo database.
In general, the storage of an integer needs to occupy 4 bytes, namely, 32 bits and N integers, then the storage space of 32N bits is needed, and whether a certain integer exists is queried, N integers need to be traversed to judge, and the time complexity is O (N); in the storage mode of this embodiment, one bit is used to represent an integer, the storage of N integers only needs to occupy N bit memory spaces, the memory becomes 1/32 of the original memory, and for the Redis database, when whether an integer exists is queried, the built-in method getbit of the Redis database can be used for judging, the time complexity is O (1), and the query speed is reduced from the original linear order to the constant order. The storage mode not only reduces the used memory space, but also greatly improves the query performance, and the results can be obtained rapidly through bit operation for the intersection and union query of a plurality of labels.
S3: grouping users according to the user data and the labels and pushing information according to the allocated user groups;
in the step of grouping the users according to the user data and the labels and pushing the information according to the allocated user groups, the user grouping mode can be combined with the actual operation condition to perform grouping. For example: pushing the coupon information aiming at the users with the consumption higher than the set threshold value and the consumption lower than the set threshold value, and monitoring the consumption condition of the users. Specifically: defining a user with more than M elements consumed daily or more than N elements consumed accumulated monthly as a VIP user; defining a user which consumes more than m-ary daily or accumulates more than n-ary consumption monthly and is a specific channel as a channel VIP user; defining the user with the last login distance of more than 7 days as a lost user, the user with the last login distance of 4 to 6 days as a pre-lost user, the user with the last login distance of 1 to 3 days as an active user, and the like.
According to the distributed user groups, different coupon information pushing is implemented, so that user consumption can be stimulated, and retention is improved, for example: when the stimulus degree of two types of coupons, namely the full coupon and the discount coupon, to the consumption of the user in marketing needs to be tested, A/B tests can be conducted on the same user group, and the stimulus degree of different coupon types to the consumption of the user is tested through test comparison, so that an optimal coupon scheme is formulated. And monitoring the consumption condition and the retention condition of the user group with the pushed coupon information.
The steps of pushing recall information aiming at the lost user and the user about to be lost and monitoring the recall condition of the lost user comprise the following steps: and acquiring information of the lost player, namely the lost player, according to the label of the lost user, and formulating recall information and pushing the recall information according to the user portrait information of the corresponding player, wherein the recall information pushing modes comprise but are not limited to short message notification, voucher issuing and return visit communication modes. And carrying out operation monitoring on the successfully awakened player in the recall of the lost user, wherein the operation monitoring comprises statistics and display of behavior events of the user, including login, search keywords, recharging, consumption and the like, and providing data visualization, data interaction and data monitoring functions for operators. In one other embodiment, the step further includes monitoring and displaying the change of the number of users before and after the information is pushed, where the number of users before and after the information is pushed includes: the number of consultation of players, the number of follow-up players, the number of recall players, the recall ratio of recall players and the like are convenient for intuitively knowing the pushing effect of the information platform.
Compared with the prior art, the game platform information pushing method provided by the invention effectively combines a large number of user data sets in the server, can quickly and accurately position users, master the user characteristics in all directions and at multiple angles, and pushes information through the distributed user groups, so that quick matching between push content and user preference is realized, the matching accuracy is improved, and network resources are saved; according to the invention, the loss probability of the user is analyzed and predicted in a fusion model mode, so that the generalization capability of the model is greatly improved, and the loss early warning accuracy of the user is high. The invention also stores the label information by using the bitmap algorithm, and compared with the traditional structured label storage scheme, the query efficiency is greatly improved under the condition of large data volume.
The invention also provides a game platform information pushing system, as shown in fig. 5, comprising:
a data acquisition module 1, configured to acquire user data in a server;
the data acquisition module 1 includes:
the server data acquisition module is used for acquiring structured and unstructured data reported to a server database and a log file reported to a server local disk;
and the data processing module is used for combining logs generated by corresponding users in the system for unstructured data, splicing the logs into an independent text corresponding to the unique identifier of the user, and cleaning the independent text by using a word segmentation tool.
The label building module 2 is used for building a user basic information label, a user consumption prediction label, a game preference label and a loss user label based on the user data; the label creation module 2 includes: the system comprises a user basic information label building module, a user consumption prediction label building module, a game preference label building module and a loss user label building module.
The user basic information label building module comprises: the registration behavior label establishing unit is used for acquiring the registration time, registration duration, registration equipment and other information of the user to establish a registration behavior label;
The user activity type label building unit is used for counting user login time points, calculating user activity time periods and building user activity type labels;
the city grade label establishing unit is used for counting the user login IP and analyzing out the user login address so as to establish the city grade label of the user;
the user activity index label building unit is used for calculating the user activity index according to the user login time and building a user activity index label;
the user consumption label establishing unit is used for establishing a user consumption behavior label and a user consumption index label according to the consumption times and the consumption amount of the user participating in the activity and the user browsing behavior data.
The user consumption prediction label establishment module comprises:
the data acquisition unit is used for acquiring and preprocessing user data from the server log;
the feature processing unit is used for extracting user basic attribute features and game behavior attribute features of paid users and unpaid users in one period and carrying out feature engineering processing on the user basic attribute features and the game behavior attribute features;
the classification model construction unit is used for constructing a classification model by utilizing a gradient lifting decision tree and adjusting parameters by a K-fold cross validation method;
The regression model feature processing unit is used for extracting user basic attributes and game behavior attribute features of the paying users in one period and carrying out feature engineering processing on the user basic attributes and the game behavior attribute features;
the regression model construction unit is used for constructing a regression model by utilizing gradient lifting regression and adjusting parameters by a K-fold cross validation method;
the user consumption rule making unit is used for making a user consumption grade rule;
and the integration unit is used for integrating the classification model, the regression model and the user consumption level rule into a user consumption prediction model.
The game preference label establishing module comprises:
the initial preference calculating unit is used for counting the number of related behavior events generated by the user before registering each type of game, and calculating to obtain the initial preference of the corresponding type of game;
a growth rate calculation unit for counting the time of the user input to each type of game in the period and the time of the user input in the previous period, and calculating the growth rate of the growth amount of each type of game time according to the time;
a preference calculating unit for calculating each type of T according to the following manner i A value;
wherein alpha is i For the increase rate of the increase amount of the game time of the user, deltax is the game time input by the user, T j Initial value, t, representing user preference i -t j Represents time interval, M represents total consumption of users in a certain game, lambda is constant, represents proportion of consumption to heat increase, T i Representing the user at t i The period favorites for a certain class of games.
And the user preference determining unit is used for sequencing according to the favorites of each game type in the user period and determining the game type preferred by the user according to the favorites value.
The churn user label building module comprises:
the characteristic data set acquisition unit is used for associating the user data with the tag by utilizing the unique user identifier to acquire a characteristic data set;
the characteristic processing unit is used for carrying out characteristic pretreatment on the characteristic data set and carrying out characteristic selection by utilizing the gradient lifting tree and the characteristic weight;
the fusion model building unit is used for training a base learning model by using a K-fold cross validation method and building a fusion model;
and the loss user label generating unit is used for calling the fusion model to identify the loss user and generating loss user label information.
The tag storage module 3 is used for taking the tag names of the user basic information tag, the user consumption prediction tag, the game preference tag and the loss user tag as keys, converting the user under the tag into a value corresponding to the key, and storing the keys and the values into the database in a mutually corresponding mode;
The tag storage module 3 includes:
the acquisition unit is used for acquiring the label name, the maximum user identification and the unique identifications of all users with the label;
the conversion unit is used for converting the unique identifications of all users in the tag into a k-bit bitmap array through a bitmap algorithm; wherein k=1+n/32, n being the number of users of the tag;
and the storage unit is used for converting the bitmap array into hexadecimal character strings, taking the label name as a key, and storing the hexadecimal character strings in the database as the value corresponding to the key.
And the information pushing module 4 is used for grouping the users according to the user data and the labels and pushing the information according to the allocated user groups.
The information push module 4 includes:
the consumption pushing unit is used for pushing the coupon information aiming at the users with consumption higher than the set threshold value and consumption lower than the set threshold value and monitoring the consumption condition of the users;
the loss recall unit is used for pushing recall information aiming at lost users and users to be lost and monitoring the number of the recall users;
the monitoring unit is used for monitoring the change of the number of users before and after information pushing and displaying the information.
Compared with the prior art, the game platform information pushing system provides a complete set of systematic module co-cooperation, builds an intelligent user operation platform, is suitable for platform operation requirements in various fields such as electronic commerce, social networks and the like, effectively combines a large number of user data sets, analyzes user characteristics in all directions and at multiple angles, realizes quick matching between pushing content and user preference by grouping users and pushing information according to the allocated user groups, improves matching accuracy and saves network resources.
The present invention also provides a computer-readable storage medium having stored thereon a computer program which, when executed by a processor, implements the steps of the game platform information pushing method as described above.
The present invention may take the form of a computer program product embodied on one or more storage media (including, but not limited to, magnetic disk storage, CD-ROM, optical storage, etc.) having program code embodied therein. Computer-readable storage media include both non-transitory and non-transitory, removable and non-removable media, and information storage may be implemented by any method or technology. The information may be computer readable instructions, data structures, modules of a program, or other data. Examples of storage media for a computer include, but are not limited to: phase change memory (PRAM), static Random Access Memory (SRAM), dynamic Random Access Memory (DRAM), other types of Random Access Memory (RAM), read Only Memory (ROM), electrically Erasable Programmable Read Only Memory (EEPROM), flash memory or other memory technology, compact disc read only memory (CD-ROM), digital Versatile Disks (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices, or any other non-transmission medium, may be used to store information that may be accessed by the computing device.
The invention also provides a computer device, which comprises a storage, a processor and a computer program stored in the storage and executable by the processor, wherein the steps of the game platform information pushing method are realized when the processor executes the computer program.
The present invention is not limited to the above-described embodiments, but, if various modifications or variations of the present invention are not departing from the spirit and scope of the present invention, the present invention is intended to include such modifications and variations as fall within the scope of the claims and the equivalents thereof.
Claims (8)
1. The game platform information pushing method is characterized by comprising the following steps of:
acquiring user data in a server;
establishing a user basic information label, a user consumption prediction label, a game preference label and a loss user label based on the user data;
grouping users according to the user data and the labels and pushing information according to the allocated user groups;
the label names of the user basic information labels, the user consumption prediction labels, the game preference labels and the loss user labels are used as keys, the users under the labels are converted into values corresponding to the keys, and the keys and the values are stored in a database in a mutually corresponding mode; wherein, this step specifically includes:
Acquiring a tag name, a maximum user identifier and unique identifiers of all users with the tag;
converting unique identifications of all users in the tag into a k-bit bitmap array through a bitmap algorithm; wherein k=1+n/32, n being the number of users of the tag;
converting the bitmap array into hexadecimal character strings, taking the label name as a key, and storing the hexadecimal character strings in a database as values corresponding to the key;
the step of establishing a game preference tag based on user data includes:
counting the number of related behavior events generated by a user before registering each type of game, and calculating to obtain the initial favorites of the corresponding type of game;
counting the time input by the user to each type of game in the current period and the time input in the previous period, and calculating the growth rate of each type of game time according to the time input by the user in the current period;
the user preference for each game type is calculated according to the following manner:
wherein alpha is i For the increase rate of the increase amount of the game time of the user, deltax is the game time input by the user, T j Initial value, t, representing user preference i -t j Represents time interval, M represents total consumption of users in a certain game, lambda is constant, represents proportion of consumption to heat increase, T i Representing the user at t i The preference of a period to a certain type of game;
and ordering the favorites of the game types in the period according to the user, and determining the game types preferred by the user according to the favorites value.
2. The game platform information pushing method according to claim 1, wherein: the step of establishing a user basic information tag based on the user data comprises the following steps:
acquiring user registration time, registration duration, registration equipment and other information to establish a registration behavior label;
counting user login IP, resolving user login address, and establishing city grade label of user;
counting user login time points, calculating user activity time periods, and establishing user activity type labels;
calculating a user activity index according to the user login time, and establishing a user activity index label;
and analyzing the browsing behaviors of the user according to the consumption times and the consumption amount of the user participating in the activities, and establishing a user consumption behavior label and a user consumption index label.
3. The game platform information pushing method according to claim 1, wherein: the step of establishing a user consumption prediction tag based on user data comprises:
collecting user data from a server log;
Extracting user basic attribute characteristics and game behavior attribute characteristics of paid users and unpaid users in a period, and carrying out feature engineering processing on the user basic attribute characteristics and the game behavior attribute characteristics;
constructing a classification model by utilizing a gradient lifting decision tree and adjusting parameters of a regression model by a K-fold cross validation method;
extracting user basic attributes and game behavior attribute characteristics of a paying user in a period, and carrying out feature engineering processing on the user basic attributes and the game behavior attribute characteristics;
constructing a regression model by utilizing gradient lifting regression and adjusting parameters of the regression model by a K-fold cross validation method;
formulating a user consumption level rule;
and integrating the classification model, the regression model and the user consumption level rule into a user consumption prediction model.
4. The game platform information pushing method according to claim 1, wherein: the step of establishing the churn user tag based on the user data comprises the following steps:
associating the user data with the tag by using the unique user identifier to obtain a characteristic data set;
performing feature selection on the feature data set by utilizing the gradient lifting tree and the feature weight;
training a plurality of basic learning models by using a K-fold cross validation method, and constructing a fusion model according to the output results of the plurality of basic learning models;
And calling the fusion model to identify the loss user and generating the label information of the loss user.
5. The game platform information pushing method according to claim 1, wherein: the step of grouping the users according to the user data and the labels and pushing the information according to the allocated user groups specifically comprises the following steps:
pushing coupon information aiming at users with consumption higher than a set threshold value and consumption lower than the set threshold value, and monitoring the consumption condition of the users;
carrying out recall information pushing aiming at lost users and users to be lost, and monitoring the number of recall users;
and monitoring and displaying the change of the number of users before and after information pushing.
6. The game platform information pushing system is characterized in that: comprising the following steps:
the data acquisition module is used for acquiring user data in the server;
the label building module is used for building a user basic information label, a user consumption prediction label, a game preference label and an attrition user label based on user data;
the information pushing module is used for grouping the users according to the user data and the labels and pushing the information according to the allocated user groups;
the label storage module is used for taking the label names of the user basic information labels, the user consumption prediction labels, the game preference labels and the loss user labels as keys, converting the user under the labels into values corresponding to the keys, and storing the keys and the values into the database in a mutually corresponding mode;
The tag storage module includes:
the acquisition unit is used for acquiring the label name, the maximum user identification and the unique identifications of all users with the label;
the conversion unit is used for converting the unique identifications of all users in the tag into a k-bit bitmap array through a bitmap algorithm; wherein k=1+n/32, n being the number of users of the tag;
the storage unit is used for converting the bitmap array into hexadecimal character strings, taking the label name as a key, and storing the hexadecimal character strings in the database as the value corresponding to the key;
the game preference label establishing module comprises:
the initial preference calculating unit is used for counting the number of related behavior events generated by the user before registering each type of game, and calculating to obtain the initial preference of the corresponding type of game;
a growth rate calculation unit for counting the time of the user input to each type of game in the current period and the time of the user input in the previous period, and calculating the growth rate of the growth amount of each type of game time according to the time;
a preference calculating unit for calculating each type of T according to the following manner i A value;
wherein alpha is i For the increase rate of the increase amount of the game time of the user, deltax is the game time input by the user, T j Initial value, t, representing user preference i -t j Represents time interval, M represents total consumption of users in a certain game, lambda is constant, represents proportion of consumption to heat increase, T i Representing the user at t i The preference of a period to a certain type of game;
and the user preference determining unit is used for sequencing according to the favorites of each game type in the user period and determining the game type preferred by the user according to the favorites value.
7. A computer-readable storage medium having stored thereon a computer program, characterized by: the computer program when executed by a processor performs the steps of the game platform information pushing method according to any of claims 1-5.
8. A computer device, characterized by: comprising a memory, a processor and a computer program stored in the memory and executable by the processor, the processor implementing the steps of the game platform information pushing method according to any of claims 1-5 when the computer program is executed.
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 CN110222267A (en) | 2019-09-10 |
| CN110222267B true 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) |
Families Citing this family (30)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| 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 |
| 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 |
| 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 |
| 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 |
| 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 |
| 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 |
| 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 |
| 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 |
| CN112494952B (en) * | 2020-09-04 | 2024-05-14 | 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 |
| 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 |
| 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 |
| CN113014476B (en) * | 2021-03-17 | 2023-04-07 | 维沃移动通信有限公司 | Group creation method and device |
| 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 |
| 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 |
| 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 |
| CN115168740B (en) * | 2022-09-06 | 2022-11-15 | 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 |
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 |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110222267A (en) | 2019-09-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110222267B (en) | Game platform information pushing method, system, storage medium and equipment | |
| Sun et al. | Discovering themes and trends in transportation research using topic modeling | |
| Oliveira et al. | Can social media reveal the preferences of voters? A comparison between sentiment analysis and traditional opinion polls | |
| US8166032B2 (en) | System and method for sentiment-based text classification and relevancy ranking | |
| CN108363821A (en) | A kind of information-pushing method, device, terminal device and storage medium | |
| CA2617954C (en) | Method and system for extracting web data | |
| US20070198459A1 (en) | System and method for online information analysis | |
| CN109002492B (en) | Performance point prediction method based on LightGBM | |
| Uddin et al. | Predicting the popularity of online news from content metadata | |
| CN114238573B (en) | Text countercheck sample-based information pushing method and device | |
| CN112749330B (en) | Information pushing method, device, computer equipment and storage medium | |
| Leydesdorff et al. | Statistical significance and effect sizes of differences among research universities at the level of nations and worldwide based on the leiden rankings | |
| CN114693409A (en) | Product matching method, device, computer equipment, storage medium and program product | |
| Kumar et al. | Fake news detection of Indian and United States election data using machine learning algorithm | |
| US10990881B1 (en) | Predictive analytics using sentence data model | |
| CN116701772A (en) | Data recommendation method and device, computer readable storage medium and electronic equipment | |
| Silva et al. | A categorical clustering of publishers for mobile performance marketing | |
| CN110717089A (en) | User behavior analysis system and method based on weblog | |
| KR101549188B1 (en) | Apparatus and method for measuring brand image | |
| Osaysa | Improving the quality of marketing analytics systems | |
| Utama et al. | SCIENTIFIC ARTICLES RECOMMENDATION SYSTEM BASED ON USER’S RELATEDNESS USING ITEM-BASED COLLABORATIVE FILTERING METHOD | |
| CN113505117A (en) | Data quality evaluation method, device, equipment and medium based on data indexes | |
| TWI419071B (en) | Active knowledge management system, method and computer program product for problem solving | |
| CN111460300A (en) | Network content pushing method and device and storage medium | |
| CN115660722B (en) | Prediction method and device for silver life customer conversion and electronic equipment |
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 |