CN117114820A - Method and system for calculating optimal push index of offline user and online user - Google Patents
Method and system for calculating optimal push index of offline user and online user Download PDFInfo
- Publication number
- CN117114820A CN117114820A CN202311371639.6A CN202311371639A CN117114820A CN 117114820 A CN117114820 A CN 117114820A CN 202311371639 A CN202311371639 A CN 202311371639A CN 117114820 A CN117114820 A CN 117114820A
- Authority
- CN
- China
- Prior art keywords
- user
- push
- pushing
- index
- information
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 29
- 230000000694 effects Effects 0.000 claims abstract description 26
- 238000004458 analytical method Methods 0.000 claims description 33
- 238000004364 calculation method Methods 0.000 claims description 11
- 230000008859 change Effects 0.000 claims description 6
- 238000007418 data mining Methods 0.000 claims description 6
- 238000005516 engineering process Methods 0.000 claims description 3
- 238000009472 formulation Methods 0.000 claims 2
- 239000000203 mixture Substances 0.000 claims 2
- 230000006399 behavior Effects 0.000 description 33
- 238000006243 chemical reaction Methods 0.000 description 14
- 238000004422 calculation algorithm Methods 0.000 description 10
- 230000004044 response Effects 0.000 description 6
- 238000012549 training Methods 0.000 description 6
- 230000002159 abnormal effect Effects 0.000 description 4
- 238000004140 cleaning Methods 0.000 description 4
- 238000000605 extraction Methods 0.000 description 4
- 238000007781 pre-processing Methods 0.000 description 4
- 239000013598 vector Substances 0.000 description 4
- 238000004891 communication Methods 0.000 description 3
- 238000011156 evaluation Methods 0.000 description 3
- 238000012360 testing method Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 2
- 238000001914 filtration Methods 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 238000005065 mining Methods 0.000 description 2
- 238000012544 monitoring process Methods 0.000 description 2
- 238000010606 normalization Methods 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 230000009467 reduction Effects 0.000 description 2
- 230000004043 responsiveness Effects 0.000 description 2
- 230000003442 weekly effect Effects 0.000 description 2
- 238000012300 Sequence Analysis Methods 0.000 description 1
- 238000002790 cross-validation Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000000611 regression analysis Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
- 238000012800 visualization Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/06—Buying, selling or leasing transactions
- G06Q30/0601—Electronic shopping [e-shopping]
- G06Q30/0631—Item recommendations
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/10—Pre-processing; Data cleansing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
- G06F18/2135—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods based on approximation criteria, e.g. principal component analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- 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)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Evolutionary Biology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Business, Economics & Management (AREA)
- Finance (AREA)
- Accounting & Taxation (AREA)
- Development Economics (AREA)
- Strategic Management (AREA)
- Economics (AREA)
- Marketing (AREA)
- General Business, Economics & Management (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a method and a system for calculating an optimal push index of an offline user and an online user, and relates to the technical field of information push. The method comprises the following steps: collecting user information; analyzing the user information to obtain the characteristic information of the user, and establishing a user portrait according to the characteristic information; determining a push index range of push content, thereby determining a push strategy; calculating push indexes of the offline user and the online user according to the push index range, and determining push priority according to the push index of the offline user and the push index of the online user; executing a pushing task according to preset pushing frequency and priority; s6, adaptively adjusting the pushing index and the pushing strategy; according to the method and the device, different pushing strategies can be formulated for different users by combining the user behavior data and the information, so that the purchase and browsing rate of the users are improved, meanwhile, the optimal pushing index can be obtained through repeated adjustment, and the recommendation effect of the APP of the electronic commerce is improved.
Description
Technical Field
The invention relates to the technical field of information push, in particular to a method and a system for calculating optimal push indexes of an offline user and an online user.
Background
In recent years, with the rise of the e-commerce APP, a recommendation system gradually becomes an important component of the e-commerce APP. The recommendation system of the e-commerce APP generally adopts a plurality of algorithms such as collaborative filtering, content filtering and the like, and recommends related goods or services to the user by analyzing information such as user behaviors, commodity attributes, user demographics and the like.
However, the current recommendation system of the e-commerce APP has many problems, such as poor recommendation effect, inaccurate recommendation result, and the like. Therefore, how to improve the accuracy and effectiveness of recommendation by analyzing the user information and behaviors is a problem to be solved by the e-commerce APP recommendation system.
Disclosure of Invention
The invention aims to provide a method and a system for calculating the optimal push index of an offline user and an online user, which combine user behavior data and information to formulate different push strategies for different users, so that the purchase and browsing rate of the users are improved, and meanwhile, the optimal push index can be obtained through repeated adjustment, and the recommendation effect of an e-commerce APP is improved.
In order to achieve the above purpose, the present invention discloses a method for calculating an optimal push index of an offline user and an online user, which comprises the following steps:
s1, collecting user information;
s2, establishing a user portrait, analyzing the user information through a data mining technology to obtain the characteristic information of the user, and establishing the user portrait according to the characteristic information;
s3, formulating a push strategy, analyzing push content according to the user portrait, and determining a push index range of the push content so as to determine the push strategy;
s4, calculating push indexes of an offline user and an online user according to the push index range, and determining push priority according to the push index size of the offline user and the push index size of the online user;
s5, pushing is executed, and pushing tasks are executed according to a preset pushing frequency and the priority according to a calculation result so as to obtain an optimal pushing effect;
s6, self-adaptive adjustment is carried out on the pushing index and the pushing strategy according to user feedback and user behavior change, so that the pushing effect is continuously optimized and improved.
Preferably, the step S2 specifically includes:
dividing user information into a plurality of different user groups;
and (3) independently analyzing each user group to obtain user portraits of different user groups.
Preferably, the step S3 specifically includes:
and combining the user figures, and determining the push index range of the corresponding user group according to the user behavior habit and the responsivity so as to determine the push strategy.
Preferably, the step S3 further includes:
and combining the user figures, and determining the push index range of the corresponding user group according to the push purpose, thereby determining the push strategy.
Preferably, the user information includes user basic information, behavior preference information, historical purchase record information, merchandise browsing record information, search history information, geographical location information, and usage device information.
Preferably, the characteristic information includes interest preferences, consumption capabilities and shopping frequency.
Correspondingly, the invention also discloses an optimal push index calculation system of the offline user and the online user, which comprises the following steps:
the user information acquisition unit is configured to acquire user information;
a user portrait creation unit configured to create a user portrait, analyze the user information by a data mining technique to obtain feature information of a user, and create a user portrait based on the feature information;
the pushing strategy making unit is configured to make a pushing strategy, analyze pushing contents according to the user portrait, and determine a pushing index range of the pushing contents so as to determine the pushing strategy;
the pushing index calculating unit is configured to calculate pushing indexes, calculate pushing indexes of the offline user and the online user according to the pushing index range, and determine pushing priority according to the pushing index size of the offline user and the pushing index size of the online user;
the pushing task execution unit is configured to execute pushing, and execute pushing tasks according to a preset pushing frequency and the priority according to a calculation result so as to obtain an optimal pushing effect;
the pushing tracking analysis unit is configured to adaptively adjust, and adaptively adjust the pushing index and the pushing strategy according to user feedback and user behavior change so as to continuously optimize and promote the pushing effect.
Preferably, the user portrait creation unit includes:
a splitting unit configured to split user information into a number of different user groups;
and the independent analysis unit is configured to perform independent analysis on each user group so as to obtain user portraits of different user groups.
Preferably, the push strategy making unit is configured to combine the user portraits, and determine the push index range of the corresponding user group according to the user behavior habit and the responsiveness, so as to determine the push strategy.
Preferably, the push policy making unit is further configured to combine the user portraits and determine a push index range of the corresponding user group according to the push purpose, thereby determining the push policy.
Compared with the prior art, the method and the device have the advantages that by combining the user behavior data and the information, different pushing strategies can be formulated for different users, so that the purchase and browsing rate of the users are improved, meanwhile, the optimal pushing index can be obtained through repeated adjustment, and the recommendation effect of the e-commerce APP is improved.
Drawings
FIG. 1 is a flow chart of an optimal push index calculation method for offline users and online users of the present invention;
FIG. 2 is a schematic diagram of the structural relationships of the user analysis system, the scene analysis system, the computing system, and the push system of the present invention;
FIG. 3 is a block diagram of the architecture of the optimal push index computing system of the offline user and online user of the present invention.
Detailed Description
In order to describe the technical content, the constructional features, the achieved objects and effects of the present invention in detail, the following description is made in connection with the embodiments and the accompanying drawings.
Referring to fig. 1, the method for calculating the optimal push index of the offline user and the online user in this embodiment includes the following steps:
s1, collecting user information;
s2, establishing a user portrait, analyzing the user information through a data mining technology to obtain the characteristic information of the user, and establishing the user portrait according to the characteristic information;
s3, formulating a push strategy, analyzing push content according to the user portrait, and determining a push index range of the push content so as to determine the push strategy;
s4, calculating push indexes of an offline user and an online user according to the push index range, and determining push priority according to the push index size of the offline user and the push index size of the online user;
s5, pushing is executed, and pushing tasks are executed according to a preset pushing frequency and the priority according to a calculation result so as to obtain an optimal pushing effect;
s6, self-adaptive adjustment is carried out on the pushing index and the pushing strategy according to user feedback and user behavior change, so that the pushing effect is continuously optimized and improved.
Preferably, the step S2 specifically includes:
dividing user information into a plurality of different user groups;
and (3) independently analyzing each user group to obtain user portraits of different user groups.
Preferably, the step S3 specifically includes:
and combining the user figures, and determining the push index range of the corresponding user group according to the user behavior habit and the responsivity so as to determine the push strategy.
Preferably, the step S3 further includes:
and combining the user figures, and determining the push index range of the corresponding user group according to the push purpose, thereby determining the push strategy.
Preferably, the user information includes user basic information, behavior preference information, historical purchase record information, merchandise browsing record information, search history information, geographical location information, and usage device information.
Preferably, the characteristic information includes interest preferences, consumption capabilities and shopping frequency.
Referring to fig. 1 and fig. 2, the execution process of the method for calculating the optimal push index of the offline user and the online user in this embodiment relates to a user analysis system, a scene analysis system, a computing system and a push system, and the four systems can be implemented by using a computer. Specifically, by introducing the recommendation system provided by the invention into the e-commerce APP, user behavior data is collected, processed and analyzed, so that the push strategy of the offline user and the online user is realized, and the commodity or category of interest to the user is recommended in real time. And calculating the index of the pushing strategy and the click rate according to the offline and online recommendation results and the user feedback information by recording the recommendation results and the feedback information of the user in real time. And (5) obtaining the optimal pushing strategy and commodity through repeated adjustment. The user analysis system, the scene analysis system, the computing system, and the pushing system of this embodiment are described below, respectively.
1. User analysis system
1. Collecting user information, user equipment information and user behavior data, wherein the user information comprises user ID, age, gender, interests and the like; the user equipment information comprises equipment model, an operating system and the like; the user behavior data includes user browsing web page records, purchasing behavior, click events, and the like.
2. The data need to be cleaned and preprocessed before the feature value extraction can be performed. The cleaning data may remove missing values, outliers, repeated values, and the like. The preprocessing data can perform operations such as normalization, standardization, dimension reduction and the like on the data so as to improve the accuracy and efficiency of feature extraction and classification.
3. After data preprocessing, it is necessary to select features related to the user and extract these features as feature vectors of the user. The feature selection adopts a correlation coefficient analysis correlation and Filter method. The feature extraction adopts a PCA dimension reduction method to retain main data features, reduce noise and redundant information, reduce data dimension and improve algorithm efficiency.
4. After extracting the user features, the user features are normalized and normalized by adopting a minimum maximum normalization method so as to eliminate the differences between different features and samples, thereby facilitating comparison and classification.
5. After the feature extraction is finished, feature combination is carried out by adopting a PCA algorithm so as to improve the expressive force and the classification effect of the features.
Finally, extracting, converting and combining the characteristic values of the user through the steps to establish the user portrait.
2. Scene analysis system
1. Push time analysis:
first, relevant user behavior data needs to be collected, including daily, weekly, monthly active times of the user, push records, user conversion situations, etc. And then preprocessing the data to remove abnormal values and repeated data, and changing the time data into a time sequence format.
The objective indicators of push time mainly comprise push conversion rate and conversion time. The push conversion rate can be calculated by dividing the conversion times by the push times, and the conversion time can be calculated by subtracting the push time point from the conversion time point.
Subjective indicators of push time mainly comprise user online and liveness conditions, user portraits, interest preferences and the like. The online and liveness conditions of different user groups in different time periods, and the hobbies of different user groups can be identified through analyzing the user behavior data.
By means of time sequence analysis of a statistical method, a prediction model of push time can be established based on objective and subjective indexes.
After the prediction model is established, the model is optimized in a prediction error evaluation mode, so that the accuracy and stability of the model are ensured.
And selecting the optimal pushing time according to the result of the model prediction.
2. Pushing frequency analysis:
first, relevant user behavior data needs to be collected, which includes the number of daily, weekly, monthly pushes of the user, user conversion conditions, and the like. And then preprocessing the data to remove abnormal values and repeated data, and changing the pushing frequency data into a time sequence format.
The objective index of the push frequency mainly comprises push conversion rate and push interval time. The push conversion rate can be calculated by dividing the conversion times by the push times, and the push interval time can be calculated by using the time difference between push time points.
The subjective indexes of the pushing frequency mainly comprise the receiving degree of the pushing of the user, the fatigue of the user, the portrait of the user and the like. The acceptance degree and fatigue degree conditions of different user groups on the pushing frequency and the hobbies of different user groups can be identified by analyzing the user behavior data.
Based on objective and subjective indexes, a prediction model of push frequency can be established. A hybrid model is built based on machine learning regression analysis.
After the prediction model is established, the model is optimized in a prediction error evaluation mode, so that the accuracy and stability of the model are ensured.
And selecting the optimal push frequency according to the result of model prediction.
3. Pushing commodity analysis:
and collecting behavior data such as browsing, clicking, purchasing, collecting and the like of a user, and simultaneously acquiring commodity information including related information such as commodity titles, descriptions, pictures, prices, sales and the like.
And cleaning abnormal data, processing the missing value, normalizing or standardizing the data, and removing redundant information and irrelevant information.
Extracting features from commodity information and user behavior data, such as browsing times, click rate, purchasing rate, collection rate and the like of a user; sales volume, price fluctuation, time to shelf, etc. of the commodity. The text information is processed, and the text is converted into feature vectors through TF-IDF and Word2 Vec.
And constructing positive and negative samples according to the user behaviors and commodity information. The positive sample is the commodity of the real purchasing behavior, and the negative sample is the commodity not purchased by the user.
The data set is divided into a training set and a test set for training the model and evaluating the model effect.
The dataset is trained by a support vector machine algorithm.
And training the training set data by using node dividing conditions, and simultaneously performing parameter tuning.
And evaluating the test set by using the accuracy, the recall and the F1 value.
And applying the trained model to actual user and commodity data to generate a push weight index of the corresponding commodity.
4. Online and offline analysis:
first, online and offline data of the user is collected, which may include information of the user's online and offline time, user behavior data, user interaction data with push messages, user equipment, and usage environment, etc.
And cleaning and arranging the collected data, removing error and repeated data, processing missing values and abnormal values, and merging, splitting and encoding the data according to the requirement.
Based on the user portrait, pushing time analysis results, pushing frequency analysis results, pushing commodity analysis results, extracting time features, user features, equipment features, frequency features and other data.
And analyzing the extracted characteristic data, exploring the relation between the characteristics and the online and offline state of the user by means of visualization, clustering and the like so as to provide basis and guidance for establishing a model.
And selecting a proper algorithm and a proper model for training according to analysis results and problem characteristics.
The trained model is evaluated by the methods of cross validation, A/B test and the like, performance indexes such as accuracy, recall rate and the like are analyzed, and the model is optimized and adjusted according to the evaluation result.
And applying the trained model to the scene of the predicted online and offline users, and pushing the proper information for the users. Predicting the possible online state of the user through a pushing time analysis model, analyzing the response of the user, and adjusting a pushing strategy and optimizing the pushing effect according to the model result.
The performance of the model in the actual scene is continuously monitored, and new data is used for training and updating the model periodically so as to keep the accuracy and effectiveness of the model.
3. Computing system
1. Calculating push index:
and defining the calculation weight of the user image, pushing time, pushing frequency and pushing commodity analysis characteristic value. In this embodiment, a linear weighting model is used to calculate the push weight, and the formula is as follows:
user portrayal feature weight = α1 age + α2 gender + α3 interest + α4 number of clicks + α4 number of purchases;
push commodity feature weight = α5 price + α6 sales + α7 score;
push time feature weight = α7 active time + α7 push conversion rate;
push frequency characteristic weight = α8 times of push + α9 times of push + α10 times of push conversion rate;
push commodity feature weight = α11 price + α12 sales + α13 score;
online and offline push weights = β1 user profile weight + β2 push time weight + β3 push frequency + β4 push commodity + β5 feedback value;
wherein α1..α13, β1..β5 are generation optimized weight parameters.
2. The weight parameters are preset, push history data of an online user and an offline user are taken to optimize the weight parameters through small-batch gradient descent, and the method comprises the following steps of:
1) Random sampling, randomly interesting a batch of data from historical data.
2) Forward propagation, for this batch of data, it is input into the model for forward propagation, and the output value for each weight is calculated.
2) Counter-propagating, calculating the gradient of each weight, and combining it into one gradient vector. Specifically, for an objective function f (x) and a batch of data, the gradient of each weight is calculated:
∇w_i=1/batch_size*∑j∈batch(∂f(x_j)/∂w_i),
where batch_size is the size of the batch, x_j is the jth data sample, and w_i is the ith weight.
Combining the gradients of all weights to obtain a gradient vector of the whole model:
g=[∇w1,∇w2,...,∇wm],
where m is the number of weights.
4) The weights are updated. The value of each weight is updated using the following formula:
w_i=w_i-θ*∇w_i,
where θ is the learning rate and ∇ w_i is the gradient of the i-th weight.
5) Repeating steps 1 to 4) until a stop condition is satisfied. Common stopping conditions include maximum number of iterations, objective function value changes not exceeding a threshold, etc.
4. Push system
1. The online pushing index and the offline pushing index are calculated through a computing system, online users and offline users are screened, users, commodities and time with high priority are obtained through the computing system for pushing users, pushing time, pushing frequency and pushing commodities, and then pushing is carried out according to the computing result.
2. Establishing a real-time monitoring system, monitoring the push success rate, push response time, click rate and loss rate of a user, and providing a user feedback inlet to acquire feedback condition of the user on push.
3. The online pushing implementation mode comprises the following steps:
1) The server establishes TCP communication connection with the APP:
TCP communication connection is established between the server and the APP, and a WebSocket mode or a long connection mode can be adopted so as to achieve pushing instantaneity and high efficiency.
2) Calling the event to realize real-time pushing:
the server side sends a message pushing event to the APP by calling an Application Programming Interface (API), and sends the pushed message to the APP side in real time.
3) Recording the push time received by a user:
after the APP receives the push message, the time stamp of the push message is recorded, and the time stamp is sent to the server.
4) Recording the time of clicking push content by a user:
when a user clicks push content, the APP sends a time stamp of the user clicking push to the server, so that the server records the time of the user clicking push.
5) Statistics of response time and user click rate:
after receiving the user's time stamp, the server can calculate the response time of the user according to the push-sent time stamp and the user's click time stamp, and count the user click rate.
6) Optimizing push algorithm and content:
according to the response time and click rate of the user, the push algorithm and push content are optimized by combining the behavior data and attribute characteristics of the user, and the push effect and the user satisfaction are improved.
In a word, the online pushing mode needs real-time communication between the server and the APP, pushing and responding are achieved through events, behavior data of users are counted, pushing effects are analyzed and estimated, pushing algorithms and content are optimized, and accordingly satisfaction and viscosity of the users are improved.
4. Offline push implementation:
and a third-party offline pushing tool is adopted, an offline user is pushed by calling a pushing API, the pushing success rate, response time and clicking times are obtained, the lost user is marked with characteristics, and a pushing algorithm and content are optimized, so that the satisfaction degree and viscosity of the user are improved. Meanwhile, the algorithm is required to be continuously optimized, analysis and mining are carried out on the data, and the pushing effect and the user satisfaction are continuously improved.
In summary, the method for calculating the optimal push index of the offline user and the online user according to the embodiment analyzes the user portrait by establishing a user analysis system; a scene analysis system is established, commodity analysis is pushed for pushing time and pushing frequency, and online users and offline analysis are performed. And (3) establishing a computing system, combining user portraits, pushing time and pushing frequency, respectively analyzing and computing pushing weight coefficients of online and offline scenes by pushed commodities, and optimizing weight parameters. And establishing a pushing system, screening the optimal user group according to the computing system, pushing the driving time, pushing the commodity frequently, and pushing the online and offline users. The online pushing system is built for real-time pushing, offline pushing is carried out through a third-party tool, online and offline pushing conditions are monitored, the computing system continuously optimizes pushing weights for historical pushing data, and the optimal pushing index computing method for offline users and online users is achieved through the collaborative work of the four systems, so that user pushing effects of the e-commerce APP are improved, and user experience and purchase conversion rate are improved. The core thought is as follows:
firstly, the invention identifies frequent item sets of users by collecting, cleaning and mining user behavior data, and establishes indexes.
Then, according to the historical behaviors and frequent item sets of the offline user, the invention establishes a corresponding pushing strategy and recommends corresponding commodities or categories to the user so as to improve the purchase and browsing rate of the user.
Meanwhile, according to the current behavior and the historical frequent item set of the online user, the method analyzes the current user and pushes the commodities or categories interested by the user in real time.
In the pushing process, the method and the device record the recommendation result and feedback information of the user in real time, and calculate the index of the offline user pushing strategy and the click rate and the index of the online user pushing strategy and the click rate according to the offline and online recommendation results and the user feedback information.
Finally, the recommendation strategy is continuously adjusted according to the index performance, so that the purchase and browsing rate of the user is improved. And the optimal pushing index is obtained through repeated adjustment, so that the recommendation effect of the e-commerce APP is improved.
Referring to fig. 3, correspondingly, the invention also discloses a system for calculating the optimal push index of the offline user and the online user, which comprises the following steps:
a user information collection unit 10 configured to collect user information;
a user portrayal creation unit 20 configured to create a user portrayal, analyze the user information by a data mining technique to obtain feature information of the user, and create the user portrayal based on the feature information;
a push policy making unit 30 configured to make a push policy, analyze push content according to the user profile, and determine a push index range of the push content, thereby determining a push policy;
a push index calculation unit 40 configured to calculate push indexes, calculate push indexes of the offline user and the online user according to the push index range, and determine push priority according to the push index size of the offline user and the push index size of the online user;
the pushing task execution unit 50 is configured to execute pushing, and execute pushing tasks according to a preset pushing frequency and the priority according to a calculation result so as to obtain an optimal pushing effect;
the push tracking analysis unit 60 is configured to adaptively adjust, according to user feedback and user behavior change, the push index and the push strategy to continuously optimize and promote the push effect.
Preferably, the user portrait creation unit includes:
a splitting unit configured to split user information into a number of different user groups;
and the independent analysis unit is configured to perform independent analysis on each user group so as to obtain user portraits of different user groups.
Preferably, the push strategy making unit is configured to combine the user portraits, and determine the push index range of the corresponding user group according to the user behavior habit and the responsiveness, so as to determine the push strategy.
Preferably, the push policy making unit is further configured to combine the user portraits and determine a push index range of the corresponding user group according to the push purpose, thereby determining the push policy.
With reference to fig. 1-3, the invention combines user behavior data and information to formulate different pushing strategies for different users, thereby improving the purchase and browsing rate of the users, and simultaneously obtaining the optimal pushing index and improving the recommendation effect of the e-commerce APP through repeated adjustment.
The foregoing description of the preferred embodiments of the present invention is not intended to limit the scope of the claims, which follow, as defined in the claims.
Claims (10)
1. The method for calculating the optimal push index of the offline user and the online user is characterized by comprising the following steps of:
s1, collecting user information;
s2, establishing a user portrait, analyzing the user information through a data mining technology to obtain the characteristic information of the user, and establishing the user portrait according to the characteristic information;
s3, formulating a push strategy, analyzing push content according to the user portrait, and determining a push index range of the push content so as to determine the push strategy;
s4, calculating push indexes of an offline user and an online user according to the push index range, and determining push priority according to the push index size of the offline user and the push index size of the online user;
s5, pushing is executed, and pushing tasks are executed according to a preset pushing frequency and the priority according to a calculation result so as to obtain an optimal pushing effect;
s6, self-adaptive adjustment is carried out on the pushing index and the pushing strategy according to user feedback and user behavior change, so that the pushing effect is continuously optimized and improved.
2. The method for calculating the optimal push index of the offline user and the online user according to claim 1, wherein the step S2 specifically includes:
dividing user information into a plurality of different user groups;
and (3) independently analyzing each user group to obtain user portraits of different user groups.
3. The method for calculating the optimal push index of the offline user and the online user according to claim 1, wherein the step S3 specifically includes:
and combining the user figures, and determining the push index range of the corresponding user group according to the user behavior habit and the responsivity so as to determine the push strategy.
4. The method for calculating optimal push index between offline user and online user according to claim 3, wherein said step S3 further comprises:
and combining the user figures, and determining the push index range of the corresponding user group according to the push purpose, thereby determining the push strategy.
5. The method for calculating optimal push index for offline users and online users according to claim 1, wherein the user information includes user basic information, behavior preference information, historical purchase record information, commodity browsing record information, search history information, geographical location information, and usage device information.
6. The method for computing optimal push index for offline users and online users according to claim 1, wherein the characteristic information comprises interest preferences, consumption capabilities and shopping frequency.
7. An optimal push index computing system for offline users and online users, comprising:
the user information acquisition unit is configured to acquire user information;
a user portrait creation unit configured to create a user portrait, analyze the user information by a data mining technique to obtain feature information of a user, and create a user portrait based on the feature information;
the pushing strategy making unit is configured to make a pushing strategy, analyze pushing contents according to the user portrait, and determine a pushing index range of the pushing contents so as to determine the pushing strategy;
the pushing index calculating unit is configured to calculate pushing indexes, calculate pushing indexes of the offline user and the online user according to the pushing index range, and determine pushing priority according to the pushing index size of the offline user and the pushing index size of the online user;
the pushing task execution unit is configured to execute pushing, and execute pushing tasks according to a preset pushing frequency and the priority according to a calculation result so as to obtain an optimal pushing effect;
the pushing tracking analysis unit is configured to adaptively adjust, and adaptively adjust the pushing index and the pushing strategy according to user feedback and user behavior change so as to continuously optimize and promote the pushing effect.
8. The system for calculating optimal push index for offline user and online user according to claim 7, wherein the user profile creation unit comprises:
a splitting unit configured to split user information into a number of different user groups;
and the independent analysis unit is configured to perform independent analysis on each user group so as to obtain user portraits of different user groups.
9. The system for computing optimal push index for offline users and online users according to claim 7, wherein the push policy formulation unit is configured to combine user portraits and determine the push index ranges of the corresponding user groups according to user behavior habits and responsivity, thereby determining the push policy.
10. The system for computing optimal push index for offline users and online users according to claim 9, wherein the push policy formulation unit is further configured to combine user portraits and determine push index ranges for corresponding user groups based on push purposes, thereby determining push policies.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202311371639.6A CN117114820A (en) | 2023-10-23 | 2023-10-23 | Method and system for calculating optimal push index of offline user and online user |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202311371639.6A CN117114820A (en) | 2023-10-23 | 2023-10-23 | Method and system for calculating optimal push index of offline user and online user |
Publications (1)
Publication Number | Publication Date |
---|---|
CN117114820A true CN117114820A (en) | 2023-11-24 |
Family
ID=88795096
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202311371639.6A Pending CN117114820A (en) | 2023-10-23 | 2023-10-23 | Method and system for calculating optimal push index of offline user and online user |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN117114820A (en) |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104850662A (en) * | 2015-06-08 | 2015-08-19 | 浙江每日互动网络科技有限公司 | User portrait based mobile terminal intelligent message pushing method, server and system |
CN109767255A (en) * | 2018-12-06 | 2019-05-17 | 东莞团贷网互联网科技服务有限公司 | A method of it is modeled by big data and realizes intelligence operation and precision marketing |
CN110472145A (en) * | 2019-07-25 | 2019-11-19 | 维沃移动通信有限公司 | A kind of content recommendation method and electronic equipment |
CN116862592A (en) * | 2023-07-31 | 2023-10-10 | 广州有机云计算有限责任公司 | Automatic push method for SOP private marketing information based on user behavior |
-
2023
- 2023-10-23 CN CN202311371639.6A patent/CN117114820A/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104850662A (en) * | 2015-06-08 | 2015-08-19 | 浙江每日互动网络科技有限公司 | User portrait based mobile terminal intelligent message pushing method, server and system |
CN109767255A (en) * | 2018-12-06 | 2019-05-17 | 东莞团贷网互联网科技服务有限公司 | A method of it is modeled by big data and realizes intelligence operation and precision marketing |
CN110472145A (en) * | 2019-07-25 | 2019-11-19 | 维沃移动通信有限公司 | A kind of content recommendation method and electronic equipment |
CN116862592A (en) * | 2023-07-31 | 2023-10-10 | 广州有机云计算有限责任公司 | Automatic push method for SOP private marketing information based on user behavior |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106485562B (en) | Commodity information recommendation method and system based on user historical behaviors | |
US10417650B1 (en) | Distributed and automated system for predicting customer lifetime value | |
CN105183727A (en) | Method and system for recommending book | |
CN108665311B (en) | Electric commercial user time-varying feature similarity calculation recommendation method based on deep neural network | |
CN103164804A (en) | Personalized method and personalized device of information push | |
CN110490683B (en) | Offline collaborative multi-model hybrid recommendation method and system | |
CN113469730A (en) | Customer repurchase prediction method and device based on RF-LightGBM fusion model under non-contract scene | |
CN112085549A (en) | Commodity recommendation method for E-commerce platform based on data processing technology | |
CN116205675B (en) | Data acquisition method and device based on thread division | |
Orogun et al. | Predicting consumer behaviour in digital market: a machine learning approach | |
CN110880127A (en) | Consumption level prediction method and device, electronic equipment and storage medium | |
US11704540B1 (en) | Systems and methods for responding to predicted events in time-series data using synthetic profiles created by artificial intelligence models trained on non-homogenous time series-data | |
CN115760202A (en) | Product operation management system and method based on artificial intelligence | |
CN116861076A (en) | Sequence recommendation method and device based on user popularity preference | |
Mehta et al. | Collaborative personalized web recommender system using entropy based similarity measure | |
US20240202588A1 (en) | Systems and methods for cohort-based predictions in clustered time-series data in order to detect significant rate-of-change events | |
CN117271905B (en) | Crowd image-based lateral demand analysis method and system | |
CN116320626B (en) | Method and system for calculating live broadcast heat of electronic commerce | |
CN117333262A (en) | Big data demand prediction method and system based on artificial intelligence for electronic commerce platform | |
CN117237038A (en) | Commodity accurate exposure processing system based on flow engine | |
CN117131261A (en) | E-commerce content recommendation system based on AI and big data analysis | |
JP4777941B2 (en) | Demand forecasting method and apparatus | |
CN117114820A (en) | Method and system for calculating optimal push index of offline user and online user | |
CN114817741A (en) | Financial product accurate recommendation method and device | |
CN112308419A (en) | Data processing method, device, equipment and computer storage medium |
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 |