CN108763502A - Information recommendation method and system - Google Patents
Information recommendation method and system Download PDFInfo
- Publication number
- CN108763502A CN108763502A CN201810538924.5A CN201810538924A CN108763502A CN 108763502 A CN108763502 A CN 108763502A CN 201810538924 A CN201810538924 A CN 201810538924A CN 108763502 A CN108763502 A CN 108763502A
- Authority
- CN
- China
- Prior art keywords
- real
- recommendation
- model
- time
- page
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Information Transfer Between Computers (AREA)
Abstract
The invention discloses a kind of information recommendation methods, this method includes that the operation data of terminal acquisition user is sent to server, server is according to the operation data in preset time period, obtain the behavioural characteristic of user, server builds the real-time recommendation model of each application scenarios corresponding with behavioural characteristic in real time, and sends real-time recommendation model to terminal, when a triggering condition is met, terminal calls the real-time recommendation model that the feature with current application scene matches, and obtains recommendation information and shows recommendation information.The invention also discloses a kind of information recommendation system, it can be improved and recommend accuracy, it can be achieved that the recommendation based on individual subscriber.
Description
Technical field
The invention belongs to field of terminal technology more particularly to a kind of information recommendation method and systems.
Background technology
As the big data epoch arrive, major application shop or application market all introduce the personalization based on big data successively
Proposed algorithm promotes user experience.But since technology and cost limit, what these schemes mostly utilized is offline mining algorithm
The user of generation attribute of drawing a portrait is recommended, such as the features such as according to the gender of user, age, educational background, occupation, income, Xiang Yong
Recommend application in family.
The defect of the above-mentioned prior art is that user's representation data is substantially daily grade time granularity update, can only be reflected
The Long-term Interest of user, and a large amount of information requirement of user has instant, of short duration, fast-changing feature, causes to push away
Recommend inaccuracy；Be limited to privacy of user protection and certain attribute datas itself collect the influences of the factors such as difficulty, data it is accurate
Rate and coverage rate are all difficult to be promoted, and cause to recommend inaccurate；It can only be refine to the recommendation to a kind of people, cannot be accurate to for a
The recommendation of people recommends personalised effects poor；Because can not historical data newly not generated user, recommend to be not suitable for new user group,
Applicability is insufficient.
Invention content
A kind of information recommendation method of offer of the embodiment of the present invention and system, can solve inspection application content on boostrap,
It is caused to find that the problematic application content time is too long, and the application problem that download is low, liveness is low.
A kind of information recommendation method provided in an embodiment of the present invention, including：
The operation data of terminal acquisition user is sent to server；
The server obtains the behavioural characteristic of the user according to the operation data in preset time period；
The real-time recommendation model of the server corresponding with the behavioural characteristic each application scenarios of structure in real time, and to institute
It states terminal and sends the real-time recommendation model；
When a triggering condition is met, the terminal calls the real-time recommendation mould that the feature with current application scene matches
Type obtains recommendation information and shows the recommendation information.
A kind of information recommendation system provided in an embodiment of the present invention, including：
Terminal and server；
The terminal, the operation data for acquiring user are sent to the server；
The server, for according to the operation data in preset time period, obtaining the behavioural characteristic of the user；
The server, the real-time recommendation model for building each application scenarios corresponding with the behavioural characteristic in real time,
And send the real-time recommendation model to the terminal；
The terminal is used to when a triggering condition is met, call and what the feature of current application scene matched pushes away in real time
Model is recommended, recommendation information is obtained and shows the recommendation information.
From the embodiments of the present invention it is found that terminal acquires the operation data of user by client in real time, and by the behaviour
It is sent to server as data, which obtains the behavioural characteristic of the user according to the operation data in preset time period, should
The real-time recommendation model of server construction each application scenarios corresponding with behavior feature, and send the real-time recommendation to the terminal
Model, when a triggering condition is met, the terminal call the real-time recommendation model that the feature with current application scene matches, and obtain
Recommendation information simultaneously shows the recommendation information, it can be achieved that being directed to the information recommendation of individual subscriber, improves customer experience, improves information and push away
Efficiency is recommended, recommended models are diversified and can combine, and are suitble to be applied to different recommendation application scenarios, are recycled the behaviour for obtaining user
Make data, constantly realizes the update of recommended models, realize the self-perfection of information recommendation.
Description of the drawings
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, to embodiment or will show below
There is attached drawing needed in technology description to be briefly described.
Fig. 1 is the application scenarios schematic diagram of the information recommendation method provided in the embodiment of the present invention；
Fig. 2 is the flow diagram for the information recommendation method that first embodiment of the invention provides；
Fig. 3 is the flow diagram for the information recommendation method that second embodiment of the invention provides；
Fig. 4 (a), Fig. 4 (b), the example interface that Fig. 4 (c) is information recommendation method provided in an embodiment of the present invention show
It is intended to；
Fig. 5 is the structural schematic diagram for the information recommendation system that third embodiment of the invention provides；
Fig. 6 is the terminal structure schematic diagram that fourth embodiment of the invention provides.
Specific implementation mode
In order to make the invention's purpose, features and advantages of the invention more obvious and easy to understand, below in conjunction with the present invention
Attached drawing in embodiment, technical scheme in the embodiment of the invention is clearly and completely described, it is clear that described reality
It is only a part of the embodiment of the present invention to apply example, and not all embodiments.Based on the embodiments of the present invention, people in the art
The every other embodiment that member is obtained without making creative work, shall fall within the protection scope of the present invention.
Information recommendation method provided in an embodiment of the present invention can be applicable to e-commerce, news, the homepage of reading, search
Page can also be applied under the application scenarios such as the homepage of application market or application shop, software page, game page.User browsing,
Click, download either search for any one application, commodity, news either after article by pull down refresh or page layout switch i.e.
It can experience and operate relevant real time individual recommendation results just now.It is right in application market or application shop with user below
Information recommendation method provided in an embodiment of the present invention is described for the operation data of application.
Fig. 1 is the typical case schematic diagram of a scenario of information recommendation method in various embodiments of the present invention, mobile terminal 10 and clothes
Device 20 be engaged in consolidated network, mobile terminal 10 is used by built-in client (as using store, application market) acquisition in real time
The operation data at family, the specific operation data for acquiring user under each application scenarios of client, the operation data type of acquisition
Including exposure data, search data, click data, downloading data etc., the operation data of acquisition specifically includes：User identifier, behaviour
Make a check mark, application identities, content identification, operating time etc., specifically, user identifier, operation mark, application identities, content mark
Knowledge can be respectively User ID (identity), operate the content ID of ID, operation object ID, operation object.Collected operation
Data are sent to the processing that next step is carried out in server 20 in the form of single or a plurality of packing.
Application scenarios refer to the application scenarios such as homepage, software page, game page, the search page of client.
Server 20 is distributed server cluster, including access server and service server, and access server connects
Enter layer, the operation data for receiving mobile terminal reporting, access server receives the operation data of mobile terminal reporting, then
According to preset distribution rules, operation data is disaggregatedly distributed to corresponding service server and is handled.
Real-time streaming computing system is provided in server 20 to carry out in real time the 10000000000 rank data volumes that client reports
Cleaning and calculating.Specifically, after operation data of the real-time streaming computing system by access layer reception client real-time report, according to
Operation data type carries out distributed statistics, i.e., respectively by differences such as exposure data, search data, click data, downloading datas
The operation data of type goes out the behavior of user by last collect statistics by different classifications by different server statistics
Feature, behavioural characteristic are the index for evaluating user behavior feature, such as hits, download number, download conversion ratio in real time
(RCVR, Realtime Click Value Rate) etc., it is in the time window light exposure and download that this downloads conversion ratio in real time
The ratio of amount.The corresponding statistical correlation data of the user are obtained with specific reference to the User ID of the user, operation ID, application ID etc..
Real-time streaming computing system preserves the real time operating data of one section of recent duration of user using sliding time window, and by fixed
When the triggering modes such as device or data threshold, real time operating data is constantly processed into behavioural characteristic, be sent to real-time model structure
System.
Off-line model system is set in server 20, multiple off-line data models are stored in off-line model system, these
Off-line data model is generated by traditional data mining algorithm.Off-line data model includes the scale model of application, answers
Correlation model, the scale model of content, the correlation model of content, user medium-term and long-term interest model etc., model refers to leading to
Cross the processing logic of data mining algorithm generation.
Wherein, the scale model of application refers to the model for recalling similar application, i.e., when an application operating of user couple,
It is not recommended using the application of same type with this.Such as, user clicks the first news APP, then the second news APP is not recommended, and is answered
Scale model is used to exclude the recommendation of same type application.
The correlation model of application refers to the model for the application for recalling inner link, i.e., when application operating of user couple
When, it is not recommended using associated application with this.Such as, user click the first news APP, then with the first same fortune of news APP
The first reading APP of battalion quotient is not recommended, and the correlation model of application is used to exclude the recommendation of associated application.
The scale model of content refers to the model for recalling similar content, i.e., when the content operation of an application of user couple,
It is not recommended with the content of the content same type.Such as, user searches for the first news, then not with the second news as the first news category
Recommended, the scale model of content is used to exclude the recommendation of same type content.
The correlation model of content refers to the model for the content for recalling inner link, i.e., when in application of user couple
When holding operation, content associated with the content is not recommended.Such as, user search for the first books, then with the same work of the first books
The second books of person are not recommended, and the correlation model of content is used to exclude the recommendation of associate content.
The medium-term and long-term interest model of user, refers to the medium-term and long-term interest model of a kind of user, and medium-term and long-term interest model is applicable in
In the processing of more stable medium-term and long-term operation data, it is not suitable for processing real time operating data, therefore, the middle term interest of user
Model is used to exclude the recommendation of the behavioural characteristic of user.
Real-time model builds system for different real-time different with the feature construction of application scenarios of behavioural characteristic of user
Recommended models.Application scenarios refer in scenes such as the homepage of application market or application shop, search page, software page, game pages.It answers
Refer to the homepage of application market or application shop, search page, software page, the game respective constitutive characteristic of page with the feature of scene.
The feature of multiple application scenarios and each real-time recommendation model are stored in real-time model structure system.
Specifically, multiple recommended models are stored in real-time model structure system, mainly there is the single act model, multirow to be
Model, itself recalls model, off-line model combination real-time model, content correlation model, the similar mould of content at exposure negative feedback model
Type breaks up model etc. using correlation model, using scale model, behavior.
The recommended models of the above real-time model structure system will not usually be used alone, but by way of permutation and combination
Combine and plays a role jointly.Such as more behavior model hybrid-sortings can be first used, then reuse exposure negative feedback model
Into rearrangement, the advantages of having the function that while retaining two models.In addition model can according to data characteristics parallel expansion, and
It is not limited to listed above.
Real-time model builds current application scene and real-time behavioural characteristic of the system according to active user, builds real-time recommendation
Model, the real-time recommendation model are the recommendation logic to be obtained according to user's current behavior feature, current application scene.Real-time mould
Type builds system and the real-time recommendation model of structure is sent to mobile terminal 10, and mobile terminal 10 calls corresponding with current scene
Real-time recommendation model handles alternate data according to the real-time recommendation model, obtains the application that be finally presented to user
Recommendation information can be showed by recommendation in the form of arranging list.Different scenes can select not according to scene characteristic
Same real-time recommendation model is recommended with achieving the effect that maximize.
The terminal in information recommendation method in the present embodiment may include mobile phone, tablet computer, PC (PC,
Personal computer) machine etc..Specific technical solution referring to following each embodiments description, for convenience of description, following reality
The terminal in example is applied by taking mobile terminal as an example.
Fig. 2 is referred to, Fig. 2 is the flow diagram for the information recommendation method that first embodiment of the invention provides, this implementation
In example, by taking mobile phone as an example, this method includes terminal：
The operation data that S101, terminal acquire user is sent to server；
Specifically, mobile phone acquires what the operation that user carries out in preset information displayed page generated by client in real time
Data, the information displayed page refer to the operation pages of application, can also be the e-commerce page, news page under other scenes
Face, reading page etc..It is included in the information displayed page content input by user in operation data.Such as it is inputted in news pages
Search term.
Operation data type include in exposing operation, search operation, clicking operation and down operation any one or
Multiple, operation data specifically includes user identifier, behavior mark, operated object mark, the content identification of operation object and behaviour
Make the time.The wherein content identification of user identifier, behavior mark, operated object mark, operation object, is specifically as follows use respectively
The content ID of family ID, behavior ID, operation object ID, operation object, user identifier, behavior mark, operated object mark, operation pair
The content identification of elephant all has unique mark, and operation object is for example applied, the content such as application content of operation object.
Collected data are sent to server in the form of single or a plurality of packing, the real-time streaming meter in the server
Calculation system can handle collected data.
S102, the server obtain the behavioural characteristic of the user according to the operation data in preset time period；
What the preset time period can be determined by server by sliding time window, server obtains in the preset time period
The user's operation data that report of mobile phone, and according to the operation data, obtain the behavioural characteristic of the user.
Behavioural characteristic is the index for evaluating user behavior feature, such as hits, downloads number, RCVR etc..
The real-time recommendation model of S103, server corresponding with behavior feature each application scenarios of structure in real time, and to this
Terminal sends the real-time recommendation model；
It is stored with various recommended models in the server, model is to handle logic, and recommended models are recommendation information
Logic.
Server builds the real-time recommendation model of each application scenarios corresponding with behavior feature, the real-time recommendation in real time
Logic can be the combination of multiple recommended models, the real-time recommendation model of each application scenarios of structure be preserved, concurrently
Give the terminal.
The real-time recommendation model can also be built by terminal.
S104, when a triggering condition is met, this which calls and the feature of current application scene matches push away in real time
Model is recommended, recommendation information is obtained and shows the recommendation information.
The trigger condition can detect page page turning or page furbishing, and page furbishing includes the refreshing behaviour of user's active
Make or reach refreshing passive when presetting the refresh cycle, page page turning includes that page turning is preset in the page turn over operation of user's active or arrival
Passive page turning when the period may be arranged as other operations and be used as trigger condition.
When a triggering condition is met, mobile phone calls the real-time recommendation model that the feature with current application scene matches, place
Manage alternate data, obtain recommendation information, the recommendation information be it is a kind of having the successively information that puts in order, such as recommendation list, and press
The recommendation information is shown according to recommendation order.
Alternate data refers to the initial data in system corresponding with user's current operation, which can be hand
What machine was asked from server.For example, when user's operation is the search " wechat " in application market, then alternate data is that this is answered
With all wechat related datas in market.
In the embodiment of the present invention, terminal acquires the operation data of user by client in real time, and the operation data is sent out
Server is given, which obtains the behavioural characteristic of the user according to the operation data in preset time period, the server structure
The real-time recommendation model of each application scenarios corresponding with behavior feature is built, and the real-time recommendation model is sent to the terminal, it should
Terminal calls the real-time recommendation model that the feature with current application scene matches, and obtains recommendation information and shows the recommendation
For breath, it can be achieved that being directed to the information recommendation of individual subscriber, raising customer experience improves information recommendation efficiency, recommended models diversification
And can combine, it is suitble to be applied to different recommendation application scenarios, is recycled the operation data for obtaining user, constantly realizes and recommend mould
The self-perfection of information recommendation is realized in the update of type.
Fig. 3 is referred to, Fig. 3 is the flow diagram for the information recommendation method that second embodiment of the invention provides, this implementation
In example, by taking mobile phone as an example, this method includes terminal：
The operation data that S201, terminal acquire user is sent to server；
Operation data includes any one or more in exposing operation, search operation, clicking operation and down operation
All data generated are operated, operation data also specifically includes user identifier, behavior mark, operated object mark, operation object
Content identification and the operating time.
S202, the server obtain the behavioural characteristic of the user according to the operation data in preset time period；
What the preset time period can be determined by server by sliding time window, server obtains in the preset time period
The user's operation data that report of mobile phone, and according to the operation data, obtain the behavioural characteristic of the user.
Behavioural characteristic is the index for evaluating user behavior feature, such as hits, downloads number, RCVR etc..
By the operation data in preset time period, distributed statistics, statistical are carried out according to the operation data type respectively
To be distinguished according to the user identifier of each operation data type, the operation mark, the operated object mark, the time window
Statistics, the statistical result for summarizing each data type obtain the behavioural characteristic of the user, such as click volume, download, in real time download
Conversion ratio etc..
In S203, the preset recommended models stored in the server, each application with behavior characteristic matching is obtained
The recommended models of the recommended models of scene, each application scenarios being matched to are sent as real-time recommendation model to the terminal；
It is stored with various recommended models in the server, model is to handle logic, and recommended models are recommendation information
Logic.
Recommended models specifically may include：Single act model, more behavior models, exposure negative feedback model, itself recall model,
Off-line model combination real-time model, content correlation model, content scale model, using correlation model, using scale model, behavior
Break up model etc..
Wherein, single act model refers to the place that the most frequent behavioural characteristic generated recently according only to user is oriented
Logic is managed, behavior feature becomes the foundation recommended；
More behavior models refer to the processing logic integrated a plurality of operation data that user generates recently and carry out hybrid-sorting；
Negative feedback model is exposed, refers to the place for the application that user after exposure does not have further behavior reduce recommendation
Logic is managed, for example, reducing recommendation for the application that user after exposure does not click；
Itself recalls model, refers to the processing logic that do not recommend using pressure for having the behaviors such as click, search to user；
Off-line model combination real-time model refers to the off-line model first used the alternate data of user in off-line model system
It handles, then the processing logic handled with real-time behavioural characteristic；
The scale model of application refers to the model for recalling similar application, i.e., when an application operating of user couple, is answered with this
It is not recommended with the application of same type.Such as, user clicks the first news APP, then the second news APP is not recommended, the phase of application
It is used to exclude the recommendation of same type application like model.
The correlation model of application refers to the model for the application for recalling inner link, i.e., when application operating of user couple
When, it is not recommended using associated application with this.Such as, user click the first news APP, then with the first same fortune of news APP
The first reading APP of battalion quotient is not recommended, and the correlation model of application is used to exclude the recommendation of associated application.
The scale model of content refers to the model for recalling similar content, i.e., when the content operation of an application of user couple,
It is not recommended with the content of the content same type.Such as, user searches for the first news, then not with the second news as the first news category
Recommended, the scale model of content is used to exclude the recommendation of same type content.
The correlation model of content refers to the model for the content for recalling inner link, i.e., when in application of user couple
When holding operation, content associated with the content is not recommended.Such as, user search for the first books, then with the same work of the first books
The second books of person are not recommended, and the correlation model of content is used to exclude the recommendation of associate content.
Model is broken up in behavior, refers to the operand for acquiring this user between this collected real time operating data
According to upsetting the processing logic of rear mixing according to preset rules with this collected real time operating data.For example, this quilt of user
Collected real time operating data is search " caricature ", and 1 day before collected operation data is search " Quadratic Finite Element ", before
1 hour collected operation data is that search " website of buying books ", in refresh page, is if model is broken up in the behavior of being applicable in
User displaying be mixed with " caricature ", " Quadratic Finite Element ", " website of buying books " search result, can be specifically it is preceding 3 display " caricature "
Search result, it is rear 3 display " caricature " and " Quadratic Finite Element " combinatorial search as a result, again next group 3 show " website of buying books "
The combinatorial search result of " caricature ".
The above recommended models will not usually be used alone, but combine common play by way of permutation and combination and make
With.
The real-time recommendation model of server construction each application scenarios corresponding with behavior feature, the real-time recommendation logic
It can be the combination of multiple recommended models, the real-time recommendation model of each application scenarios of structure is sent to the terminal.
Further, it is additionally provided with off-line model system in server, multiple offline numbers are stored in off-line model system
According to model, these off-line data models are generated by traditional data mining algorithm.Off-line data model includes application
Scale model, the correlation model of application, the scale model of content, the correlation model of content, user medium-term and long-term interest model etc.,
Model refers to the processing logic generated by data mining algorithm.
Wherein, the scale model of the application in off-line model, the correlation model of application, the scale model of content and content
The processing logic of model of the same name in correlation model, with above-mentioned recommended models is identical, the difference is that, it is same in off-line model
Name model is the logic for handling off-line data, content scale model, content correlation model, application in real-time model structure system
Scale model is processing real time data using correlation model.
The medium-term and long-term interest model of user, refers to the medium-term and long-term interest model of a kind of user, and medium-term and long-term interest model is applicable in
In the processing of more stable medium-term and long-term operation data, it is not suitable for processing real time operating data, therefore, the middle term interest of user
Model is used to exclude the recommendation of the behavioural characteristic of user.The medium-term and long-term interest model is the long-term action by analyzing multiple users
Feature obtains.
Off-line data model mainly has following two effects：
First, carrying out prescreening to alternate data before the calculating for carrying out real-time recommendation, alternate data refers to working as with user
Initial data in the corresponding system of preceding operation, the initial data can be that mobile phone is asked from server, for example, with
Family current operation is search " caricature " in the client, then alternate data is all caricature related datas in the client.It utilizes
Off-line data model screens out a part of data from the alternate data, reduces processing delay when real-time recommendation, is promoted and is recommended
The performance of system.
Second is that in the case where user has just opened client and do not have any behavior to feed back, it is standby from this using off-line data model
It selects and screens out a part of data in data, can also realize the personalized recommendation to the user to a certain extent.
Specifically, mobile phone is in client initial start-up, to server request of loading homepage data, due to not generating reality also
When operation data, then off-line model is confirmed as real-time recommendation model by server, and is sent to mobile phone, and mobile phone calls offline mould
Type obtains recommendation information and shows the recommendation information, which is obtained by analyzing the long-term action feature of multiple users.
S204, when a triggering condition is met, the terminal call the real-time recommendation that the feature with current application scene matches
Model obtains recommendation information and shows the recommendation information.
The trigger condition can detect page page turning or page furbishing, and page furbishing includes the refreshing behaviour of user's active
Make or reach refreshing passive when presetting the refresh cycle, page page turning includes that page turning is preset in the page turn over operation of user's active or arrival
Passive page turning when the period may be arranged as other operations and be used as trigger condition.
When a triggering condition is met, which calls the real-time recommendation mould that the feature with current application scene matches
Type handles alternate data, obtains recommendation information, the recommendation information be it is a kind of have the successively information that puts in order, such as recommendation list,
And show the recommendation information according to recommendation order.
Alternate data refers to the initial data in system corresponding with user's current operation, which can be hand
What machine was asked from server.For example, when user's operation is search " caricature " in the client, then alternate data is the client
All caricature related datas at end.
It should be noted that call the real-time recommendation model in mobile phone, obtain recommendation information and show the recommendation information it
Before, mobile phone calls preset off-line model, is pre-processed to alternate data, it is possible to reduce obtained using the real-time recommendation model
The processing time of recommendation information.
Terminal calls the real-time recommendation model that the feature with current application scene matches, and obtains recommendation information and show to be somebody's turn to do
Recommendation information can be specifically to obtain the target recommended models to match with the feature of current application scene, specifically, acquisition is worked as
The feature and the operation data before acquisition page page turning or page furbishing of preceding application scenarios, and according to the spy of current application scene
Sign confirms that current application scene, current application scene include homepage, software page, game page or the search page of active client, page
Operation data before face page turning or page furbishing includes：It browses, click, download or searches for interior in user's current application scene
Hold.From the real-time recommendation model and off-line model of storage, the target recommended models to match with current application scene are obtained.
When the target recommended models include real-time recommendation model and off-line model, mobile phone call off-line model, pair with should
The corresponding alternate data of operation data before page page turning or page furbishing is pre-processed, the alternate data namely with the page
The associated alternate data of operation data before page turning or page furbishing, for example, when the current application scenarios are client homepages, behaviour
Work is search, which is " caricature ", then alternate data is all caricature related datas of the client.Further,
Based on pretreated alternate data, the real-time recommendation model for calling the feature with current application scene to match obtains recommendation
It ceases and shows the recommendation information；
When the target recommended models include real-time recommendation model, the reality to match with the feature of current application scene is called
When recommended models, obtain recommendation information and show the recommendation information.
Further, when there are multiple real-time recommendation models to match with current application scene feature, mobile phone is pressed
According to preset calling sequence, each real-time recommendation model for calling the feature with current application scene to match successively, from alternative number
Recommendation information is obtained in.Specifically, if the real-time recommendation model that there are 4 to match with the feature of current application scene, hand
Machine calls 4 real-time recommendation models, final recommendation is obtained from alternate data successively according to preset calling sequence
Breath.
One example is as shown in figure 4, Fig. 4 (a) is the search interface that user is scanned for using mobile phone in client, Fig. 4
(b) it is to return the homepage of client after user searches for, in the head Page refresh automaticallies, Fig. 4 (c) is after displaying refreshes for use
The search result that family is recommended.
Alternatively, when the real-time recommendation model that the feature with current application scene matches is multiple, terminal is called respectively
Each real-time recommendation model to match with the feature of current application scene obtains and multiple real-time recommendation moulds from alternate data
The corresponding multigroup alternative recommendation information of type, takes the target recommendation of the preset ranking in the alternative recommendation information of each group respectively
Breath, and by multigroup target recommendation information of acquirement, be ranked up according to the priority of each real-time recommendation model, obtain recommendation
Breath.Specifically, if the real-time recommendation model that there are 4 to match with the feature of current application scene, mobile phone calls 4 to be somebody's turn to do respectively
Real-time recommendation model obtains 4 groups of alternative recommendation informations corresponding with 4 real-time recommendation models from alternate data, takes 4 respectively
The target recommendation information of preset ranking in the alternative recommendation information of group, and by 4 groups of target recommendation informations of acquirement, according to each real-time
The preset priority of recommended models is ranked up, and obtains final recommendation information.
Further, in order to which the interest closer to user is recommended, the time sequencing that each recommendation information is operated is obtained
And number adjusts the priority of each real-time recommendation model according to time sequencing and number.Wherein, time sequencing is more forward, explanation
The user the interested in this recommendation information, and number is more, illustrates that the user the interested in this recommendation information, then real-time recommendation model
Priority it is higher.Can be that weight is respectively set in time sequencing and number, such as time sequencing and the weight of number are respectively
0.5 and 0.5.According to the algorithm of setting, user is to the time sequencing of the operation of recommendation information, number and the weight of setting, adjustment
The priority of each real-time recommendation model.
The priority of each real-time recommendation model after adjustment is sent to server, server is according in default statistics duration
Adjustment number is more than in the server that preset times are real by the priority of each real-time recommendation model after the adjustment that each terminal is sent
When recommended models priority be adjusted.In order to exclude accidentalia, server can after adjustment number is more than certain number,
The priority for adjusting real-time recommendation model again improves the accuracy of adjustment.
In the embodiment of the present invention, terminal acquires the operation data of user by client in real time, and the operation data is sent out
Server is given, which obtains the behavioural characteristic of the user according to the operation data in preset time period, the server structure
The real-time recommendation model of each application scenarios corresponding with behavior feature is built, and the real-time recommendation model is sent to the terminal, when
When meeting trigger condition, which calls the real-time recommendation model that the feature with current application scene matches, and obtains recommendation
It ceases and shows the recommendation information, it can be achieved that being directed to the information recommendation of individual subscriber, improve customer experience, improve information recommendation effect
Rate, recommended models are diversified and can combine, and are suitble to be applied to different recommendation application scenarios, are recycled the operand for obtaining user
According to the self-perfection of information recommendation is realized in the constantly update of realization recommended models.
Fig. 5 is referred to, Fig. 5 is that the information recommendation system that third embodiment of the invention provides only is shown for convenience of description
With relevant part of the embodiment of the present invention.The system includes：
Terminal 301 and server 302；
Terminal 301, the operation data for acquiring user in real time are sent to server 302；
Server 302, for according to the operation data in preset time period, obtaining the behavioural characteristic of the user；
Server 302, the real-time recommendation model for building each application scenarios corresponding with behavior feature in real time, and to
Terminal 301 sends the real-time recommendation model；
Terminal 301, the real-time recommendation for when a triggering condition is met, calling the feature with current application scene to match
Model obtains recommendation information and shows the recommendation information.
Further, server 302 are additionally operable to determine the preset time period by sliding time window；
Server 302 is additionally operable to the operation data in the preset time period, is carried out respectively according to the operation data type
Distribution statistics；
Server 302 is additionally operable to collect statistics result and obtains the behavioural characteristic of the user.
Server 302 is additionally operable in client initial start-up, sends off-line model to terminal 301, the off-line model is logical
It crosses and analyzes the long-term action feature of multiple users and obtain；
Terminal 301 is additionally operable to call the off-line model, obtains recommendation information and shows the recommendation information.
Terminal 301 is additionally operable to when detecting page page turning or page furbishing, obtains the feature phase with current application scene
Matched target recommended models；
Terminal 301 is additionally operable to obtain the feature of current application scene and obtains the behaviour before page page turning or page furbishing
Make data, and according to the feature of current application scene, confirms current application scene, current application scene includes active client
Homepage, software page, game page or search page, the operation data before page page turning or page furbishing include：User's current application field
The content for being browsed in scape, clicking, download or searching for is additionally operable to acquisition and recommends mould with the target that current application scene matches
Type.
Terminal 301 is additionally operable to when the target recommended models include the real-time recommendation model and off-line model, and calling should be from
Line model pre-processes the corresponding alternate data of operation data before the page page turning or page furbishing, and based on pre- place
The alternate data of reason, the real-time recommendation model for calling the feature with current application scene to match, obtains recommendation information and shows
The recommendation information；
Terminal 301 is additionally operable to when the target recommended models include the real-time recommendation model, is called and current application scene
The real-time recommendation model that matches of feature, obtain recommendation information and show the recommendation information.
Terminal 301 is additionally operable to when the real-time recommendation model that the feature with current application scene matches is multiple, according to
Preset calling sequence, each real-time recommendation model for calling the feature with current application scene to match successively, from alternate data
In obtain the recommendation information.
Terminal 301 is additionally operable to when the real-time recommendation model that the feature with current application scene matches is multiple, terminal
Call each real-time recommendation model to match with the feature of current application scene respectively, obtained from alternate data with it is multiple in real time
The corresponding multigroup alternative recommendation information of recommended models；
The target recommendation information of the preset ranking in the alternative recommendation information of each group is taken respectively, and multigroup target of acquirement is pushed away
Information is recommended, is ranked up according to the priority of each real-time recommendation model, obtains recommendation information.
Terminal 301 is additionally operable to obtain the time sequencing and number that each recommendation information is operated, and, according to time sequencing
And number, the priority of each real-time recommendation model is adjusted, and the priority of each real-time recommendation model after adjustment is sent to clothes
Business device 302；
Server 302 is additionally operable to according to each real-time recommendation mould after the adjustment that each terminal 301 is sent in default statistics duration
Adjustment number is adjusted by the priority of type in server 302 more than the priority of preset times real-time recommendation model.
Techniques not described details in the embodiment of the present invention, it is identical referring to each embodiment shown in 1~Fig. 4 of earlier figures, this
Place repeats no more.
In the embodiment of the present invention, terminal acquires the operation data of user by client in real time, and the operation data is sent out
Server is given, which obtains the behavioural characteristic of the user according to the operation data in preset time period, the server structure
The real-time recommendation model of each application scenarios corresponding with behavior feature is built, and the real-time recommendation model is sent to the terminal, when
When meeting trigger condition, which calls the real-time recommendation model that the feature with current application scene matches, and obtains recommendation
It ceases and shows the recommendation information, it can be achieved that being directed to the information recommendation of individual subscriber, improve customer experience, improve information recommendation effect
Rate, recommended models are diversified and can combine, and are suitble to be applied to different recommendation application scenarios, are recycled the operand for obtaining user
According to the self-perfection of information recommendation is realized in the constantly update of realization recommended models.
Fig. 6 is referred to, Fig. 6 is that the terminal that fourth embodiment of the invention provides illustrates only and this hair for convenience of description
The bright relevant part of embodiment.The terminal includes：
Acquisition module 401, the operation data for acquiring user；
Sending module 402, for the operation data to be sent to server；
Recommending module 403, the real-time recommendation model of each application scenarios for receiving server transmission, is triggered when meeting
When condition, the real-time recommendation model for calling the feature with current application scene to match obtains recommendation information and shows the recommendation
Information.
Further, recommending module 403 are additionally operable in client initial start-up, obtain the offline mould that server is sent
Type, which is obtained by analyzing the long-term action feature of multiple users, and calls the off-line model, obtains recommendation information
And show the recommendation information.
Terminal also one improves：
Acquisition module 404 is additionally operable to when detecting page page turning or page furbishing, obtains the spy with current application scene
Levy the target recommended models to match；
Acquisition module 404 is additionally operable to before obtaining the feature of current application scene and obtaining page page turning or page furbishing
Operation data, and according to the feature of current application scene, confirm current application scene, current application scene includes existing customer
The homepage at end, software page, game page or search page, the operation data before page page turning or page furbishing include：User currently answers
With the content for being browsed in scene, clicking, download or searching for；And it obtains the target to match with current application scene and pushes away
Recommend model.
Recommending module 403 is additionally operable to when the target recommended models include the real-time recommendation model and off-line model, is called
The off-line model, a pair alternate data corresponding with the operation data before the page page turning or page furbishing pre-processes, and base
In pretreated alternate data, the real-time recommendation model for calling the feature with current application scene to match obtains recommendation information
And show the recommendation information；
Recommending module 403 is additionally operable to when the target recommended models include the real-time recommendation model, calling and current application
The real-time recommendation model that the feature of scene matches obtains recommendation information and shows the recommendation information；
Recommending module 403 is additionally operable to when the real-time recommendation model that the feature with current application scene matches is multiple,
According to preset calling sequence, each real-time recommendation model to match with the feature of current application scene is called successively, from alternative
The recommendation information is obtained in data.
Recommending module 403 is additionally operable to call preset off-line model, be pre-processed to alternate data.
Recommending module 403 is additionally operable to when the real-time recommendation model that the feature with current application scene matches is multiple,
Terminal calls each real-time recommendation model to match with the feature of current application scene respectively, obtained from alternate data with it is multiple
The corresponding multigroup alternative recommendation information of real-time recommendation model；The mesh of the preset ranking in the alternative recommendation information of each group is taken respectively
Recommendation information is marked, and by multigroup target recommendation information of acquirement, is ranked up, obtains according to the priority of each real-time recommendation model
Recommendation information.
Acquisition module 404 is additionally operable to obtain the time sequencing and number that each recommendation information is operated；
Module 405 is adjusted, for according to time sequencing and number, adjusting the priority of each real-time recommendation model；
Sending module 402 is additionally operable to the priority of each real-time recommendation model after adjustment being sent to server.
The embodiment of the present invention does not describe to the greatest extent details, referring to the description of content same as before.
In the embodiment of the present invention, terminal acquires the operation data of user by client in real time, and the operation data is sent out
Give server so that the server obtains the behavioural characteristic of the user, and build according to the operation data in preset time period
The real-time recommendation model of each application scenarios corresponding with behavior feature, terminal receive the real-time recommendation mould of server transmission
Type, when a triggering condition is met, the terminal call the real-time recommendation model that the feature with current application scene matches, and are pushed away
It recommends information and shows the recommendation information, it can be achieved that being directed to the information recommendation of individual subscriber, improve customer experience, improve information recommendation
Efficiency, recommended models are diversified and can combine, and are suitble to be applied to different recommendation application scenarios, are recycled the operation for obtaining user
Data constantly realize the update of recommended models, realize the self-perfection of information recommendation.
The embodiment of the present application also provides a kind of computer readable storage medium, which can be
It is set in the electronic device in the various embodiments described above.It is stored with computer program on the computer readable storage medium, the journey
The information recommendation method described in earlier figures 2 and embodiment illustrated in fig. 3 is realized when sequence is executed by processor.Further, the meter
Calculation machine can storage medium can also be USB flash disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), RAM, magnetic disc
Or the various media that can store program code such as CD.
In the above-described embodiments, it all emphasizes particularly on different fields to the description of each embodiment, there is no the portion being described in detail in some embodiment
Point, it may refer to the associated description of other embodiments.
It is the description to information recommendation method provided by the present invention and information recommendation system above, for the skill of this field
Art personnel, the thought of embodiment according to the present invention, there will be changes in the specific implementation manner and application range, to sum up,
The content of the present specification should not be construed as limiting the invention.
Claims (10)
1. a kind of information recommendation method, which is characterized in that including：
The operation data of terminal acquisition user is sent to server；
The server obtains the behavioural characteristic of the user according to the operation data in preset time period；
The real-time recommendation model of the server corresponding with the behavioural characteristic each application scenarios of structure in real time, and to the end
End sends the real-time recommendation model；
When a triggering condition is met, the terminal calls the real-time recommendation model that the feature with current application scene matches, and obtains
To recommendation information and show the recommendation information.
2. the method as described in claim 1, which is characterized in that the server is according to the operand in preset time period
According to, the behavioural characteristic of the user is obtained, including：
The server determines the preset time period by sliding time window；
By the operation data in the preset time period, distributed statistics is carried out according to the operation data type respectively；
Collect statistics result obtains the behavioural characteristic of the user.
3. the method as described in claim 1, which is characterized in that the trigger condition is to detect page page turning or page brush
It is newly, described that the terminal calls the real-time recommendation model that the feature with current application scene matches when a triggering condition is met,
It obtains recommendation information and shows that the recommendation information includes：
When detecting page page turning or page furbishing, obtains the target to match with the feature of current application scene and recommend mould
Type；
When the target recommended models include the real-time recommendation model and off-line model, the terminal calls the offline mould
Type, a pair alternate data corresponding with the operation data before page page turning or page furbishing pre-process, and based on pretreated
Alternate data, the real-time recommendation model for calling the feature with current application scene to match obtain described in recommendation information and displaying
Recommendation information；
When the target recommended models include the real-time recommendation model, the feature with current application scene is called to match
Real-time recommendation model obtains recommendation information and shows the recommendation information.
4. method as claimed in claim 3, which is characterized in that the mesh that the feature of the acquisition and current application scene matches
Marking recommended models includes：
It obtains the feature of current application scene and obtains the operation data before the page page turning or page furbishing, and according to institute
It states the feature of current application scene, confirms current application scene, the current application scene includes the homepage, soft of active client
Part page, game page or search page, the operation data before the page page turning or page furbishing include：Current application field described in user
The content for being browsed in scape, clicking, download or searching for；
Obtain the target recommended models to match with the current application scene.
5. the method as described in claim 1, which is characterized in that the method further includes：
In client initial start-up, the server sends off-line model to the terminal, and the off-line model passes through analysis
The long-term action feature of multiple users obtains；
The terminal calls the off-line model, obtains recommendation information and shows the recommendation information.
6. the method as described in claim 1, which is characterized in that the terminal calls the feature with current application scene to match
Real-time recommendation model, obtaining recommendation information includes：
When the real-time recommendation model that the feature with current application scene matches is multiple, the terminal is according to preset calling
Sequentially, each real-time recommendation model for calling the feature with current application scene to match successively, obtains described from alternate data
Recommendation information.
7. the method as described in claim 1, which is characterized in that the terminal calls the feature with current application scene to match
Real-time recommendation model, obtaining recommendation information includes：
When the real-time recommendation model that the feature with current application scene matches be it is multiple when, the terminal calls respectively with currently
Each real-time recommendation model that the feature of application scenarios matches obtains and multiple real-time recommendation models point from alternate data
Not corresponding multigroup alternative recommendation information；
The target recommendation information of the preset ranking in alternative recommendation information described in each group is taken respectively, and multigroup target of acquirement is pushed away
Information is recommended, is ranked up according to the priority of each real-time recommendation model, obtains the recommendation information.
8. the method for claim 7, which is characterized in that the terminal calls the feature with current application scene to match
Real-time recommendation model, after obtaining recommendation information, further include：
Obtain the time sequencing and number that each recommendation information is operated；
According to the time sequencing and number, the priority of each real-time recommendation model is adjusted, and will be each described after adjustment
The priority of real-time recommendation model is sent to the server；
The server is according to each real-time recommendation model after the adjustment of each terminal transmission in default statistics duration
Adjustment number is adjusted by priority in the server more than the priority of preset times real-time recommendation model.
9. a kind of information recommendation system, which is characterized in that including：
Terminal and server；
The terminal, the operation data for acquiring user are sent to the server；
The server, for according to the operation data in preset time period, obtaining the behavioural characteristic of the user；
The server, the real-time recommendation model for building each application scenarios corresponding with the behavioural characteristic in real time, and to
The terminal sends the real-time recommendation model；
The terminal, the real-time recommendation mould for when a triggering condition is met, calling the feature with current application scene to match
Type obtains recommendation information and shows the recommendation information.
10. system as claimed in claim 9, which is characterized in that
The server is additionally operable to determine the preset time period by sliding time window；
The server is additionally operable to by the operation data in the preset time period, respectively according to the operation data type
Carry out distributed statistics；
The server is additionally operable to collect statistics result and obtains the behavioural characteristic of the user；
The server is additionally operable in client initial start-up, sends off-line model to the terminal, the off-line model is logical
It crosses and analyzes the long-term action feature of multiple users and obtain；
The terminal is additionally operable to call the off-line model, obtains recommendation information and shows the recommendation information；
The terminal is additionally operable to when detecting page page turning or page furbishing, obtains the feature phase with current application scene
The target recommended models matched；
The terminal is additionally operable to when the target recommended models include the real-time recommendation model and off-line model, calls institute
Off-line model is stated, a pair alternate data corresponding with the operation data before page page turning or page furbishing pre-processes, and is based on
Pretreated alternate data, the real-time recommendation model for calling the feature with current application scene to match, obtains recommendation information simultaneously
Show the recommendation information；
The terminal is additionally operable to when the target recommended models include the real-time recommendation model, is called and current application field
The real-time recommendation model that the feature of scape matches obtains recommendation information and shows the recommendation information；
The terminal is additionally operable to obtain the feature of current application scene and obtains the behaviour before page page turning or page furbishing
Make data, and according to the feature of the current application scene, confirms that current application scene, the current application scene include current
The homepage of client, software page, game page or search page, the operation data before the page page turning or page furbishing include：With
The content for being browsed in current application scene described in family, clicking, download or searching for；
Obtain the target recommended models to match with the current application scene.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810538924.5A CN108763502B (en) | 2018-05-30 | 2018-05-30 | Information recommendation method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810538924.5A CN108763502B (en) | 2018-05-30 | 2018-05-30 | Information recommendation method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108763502A true CN108763502A (en) | 2018-11-06 |
| CN108763502B CN108763502B (en) | 2022-03-25 |
Family
ID=64004084
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810538924.5A Active CN108763502B (en) | 2018-05-30 | 2018-05-30 | Information recommendation method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108763502B (en) |
Cited By (19)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110083688A (en) * | 2019-05-10 | 2019-08-02 | 北京百度网讯科技有限公司 | Search result recalls method, apparatus, server and storage medium |
| CN110516159A (en) * | 2019-08-30 | 2019-11-29 | 北京字节跳动网络技术有限公司 | A kind of information recommendation method, device, electronic equipment and storage medium |
| CN111262895A (en) * | 2018-11-30 | 2020-06-09 | 阿里巴巴集团控股有限公司 | Information processing method, system and equipment |
| CN111414540A (en) * | 2020-03-20 | 2020-07-14 | 张明 | Online learning recommendation method and device, online learning system and server |
| CN111444424A (en) * | 2020-03-25 | 2020-07-24 | 深圳市分期乐网络科技有限公司 | Information recommendation method and information recommendation system |
| CN111666485A (en) * | 2019-03-07 | 2020-09-15 | 腾讯科技（深圳）有限公司 | Information recommendation method and device and terminal |
| WO2020200146A1 (en) * | 2019-04-04 | 2020-10-08 | 阿里巴巴集团控股有限公司 | Page information processing method, device, and electronic apparatus |
| CN111797071A (en) * | 2019-04-09 | 2020-10-20 | Oppo广东移动通信有限公司 | Data processing method, data processing device, storage medium and electronic equipment |
| CN111831889A (en) * | 2019-04-15 | 2020-10-27 | 泰康保险集团股份有限公司 | Block chain-based virtual fitness application recommendation method and device |
| CN111915382A (en) * | 2019-05-08 | 2020-11-10 | 阿里巴巴集团控股有限公司 | Data processing method, system and device |
| CN112328892A (en) * | 2020-11-24 | 2021-02-05 | 北京百度网讯科技有限公司 | Information recommendation method, device, equipment and computer storage medium |
| CN112579902A (en) * | 2020-12-24 | 2021-03-30 | 第四范式（北京）技术有限公司 | Behavior data management method and device supporting multiple intelligent application scenes |
| CN113094589A (en) * | 2021-04-30 | 2021-07-09 | 中国银行股份有限公司 | Intelligent service recommendation method and device |
| CN113190758A (en) * | 2021-05-21 | 2021-07-30 | 聚好看科技股份有限公司 | Server and media asset recommendation method |
| CN113763086A (en) * | 2020-09-23 | 2021-12-07 | 北京沃东天骏信息技术有限公司 | Information recommendation method and device |
| CN113947459A (en) * | 2021-10-21 | 2022-01-18 | 北京沃东天骏信息技术有限公司 | Article pushing and selecting method and device and storage medium |
| CN114185471A (en) * | 2022-02-17 | 2022-03-15 | 哈尔滨工业大学（威海） | Clothing recommendation method based on user intention recognition |
| WO2022199494A1 (en) * | 2021-03-23 | 2022-09-29 | 华为技术有限公司 | User interest-based content recommendation method, and terminal device |
| CN116993412A (en) * | 2023-07-06 | 2023-11-03 | 道有道科技集团股份公司 | Intelligent delivery system and method based on user operation data analysis |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20020116291A1 (en) * | 2000-12-22 | 2002-08-22 | Xerox Corporation | Recommender system and method |
| US7788123B1 (en) * | 2000-06-23 | 2010-08-31 | Ekhaus Michael A | Method and system for high performance model-based personalization |
| CN103810030A (en) * | 2014-02-20 | 2014-05-21 | 北京奇虎科技有限公司 | Application recommendation method, device and system based on mobile terminal application market |
| CN103886090A (en) * | 2014-03-31 | 2014-06-25 | 北京搜狗科技发展有限公司 | Content recommendation method and device based on user favorites |
| CN104021483A (en) * | 2014-06-26 | 2014-09-03 | 陈思恩 | Recommendation method for passenger demands |
| CN104123284A (en) * | 2013-04-24 | 2014-10-29 | 华为技术有限公司 | Recommendation method and server |
| CN105183781A (en) * | 2015-08-14 | 2015-12-23 | 百度在线网络技术（北京）有限公司 | Information recommendation method and apparatus |
| CN107423442A (en) * | 2017-08-07 | 2017-12-01 | 火烈鸟网络(广州)股份有限公司 | Method and system, storage medium and computer equipment are recommended in application based on user's portrait behavioural analysis |
| US20180005297A1 (en) * | 2014-12-26 | 2018-01-04 | China Unionpay Co., Ltd. | Analysis and collection system for user interest data and method therefor |
| CN111897861A (en) * | 2020-06-30 | 2020-11-06 | 苏宁金融科技(南京)有限公司 | Content recommendation method and device, computer equipment and storage medium |
-
2018
- 2018-05-30 CN CN201810538924.5A patent/CN108763502B/en active Active
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7788123B1 (en) * | 2000-06-23 | 2010-08-31 | Ekhaus Michael A | Method and system for high performance model-based personalization |
| US20020116291A1 (en) * | 2000-12-22 | 2002-08-22 | Xerox Corporation | Recommender system and method |
| CN104123284A (en) * | 2013-04-24 | 2014-10-29 | 华为技术有限公司 | Recommendation method and server |
| CN103810030A (en) * | 2014-02-20 | 2014-05-21 | 北京奇虎科技有限公司 | Application recommendation method, device and system based on mobile terminal application market |
| CN103886090A (en) * | 2014-03-31 | 2014-06-25 | 北京搜狗科技发展有限公司 | Content recommendation method and device based on user favorites |
| CN104021483A (en) * | 2014-06-26 | 2014-09-03 | 陈思恩 | Recommendation method for passenger demands |
| US20180005297A1 (en) * | 2014-12-26 | 2018-01-04 | China Unionpay Co., Ltd. | Analysis and collection system for user interest data and method therefor |
| CN105183781A (en) * | 2015-08-14 | 2015-12-23 | 百度在线网络技术（北京）有限公司 | Information recommendation method and apparatus |
| CN107423442A (en) * | 2017-08-07 | 2017-12-01 | 火烈鸟网络(广州)股份有限公司 | Method and system, storage medium and computer equipment are recommended in application based on user's portrait behavioural analysis |
| CN111897861A (en) * | 2020-06-30 | 2020-11-06 | 苏宁金融科技(南京)有限公司 | Content recommendation method and device, computer equipment and storage medium |
Non-Patent Citations (2)
| Title |
|---|
| 孙凯涛: "基于分布式流处理框架下的移动健身管理***研究", 《现代电子技术》 * |
| 腾讯大数据: "大数据实时推荐-不只是统计", 《HTTPS://CLOUD.TENCENT.COM/DEVELOPER/ARTICLE/1030050》 * |
Cited By (26)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111262895B (en) * | 2018-11-30 | 2022-11-29 | 阿里巴巴华北技术有限公司 | Information processing method, system and equipment |
| CN111262895A (en) * | 2018-11-30 | 2020-06-09 | 阿里巴巴集团控股有限公司 | Information processing method, system and equipment |
| CN111666485B (en) * | 2019-03-07 | 2024-01-09 | 深圳市雅阅科技有限公司 | Information recommendation method, device and terminal |
| CN111666485A (en) * | 2019-03-07 | 2020-09-15 | 腾讯科技（深圳）有限公司 | Information recommendation method and device and terminal |
| WO2020200146A1 (en) * | 2019-04-04 | 2020-10-08 | 阿里巴巴集团控股有限公司 | Page information processing method, device, and electronic apparatus |
| CN111782918A (en) * | 2019-04-04 | 2020-10-16 | 阿里巴巴集团控股有限公司 | Page information processing method and device and electronic equipment |
| CN111797071A (en) * | 2019-04-09 | 2020-10-20 | Oppo广东移动通信有限公司 | Data processing method, data processing device, storage medium and electronic equipment |
| CN111831889A (en) * | 2019-04-15 | 2020-10-27 | 泰康保险集团股份有限公司 | Block chain-based virtual fitness application recommendation method and device |
| CN111915382A (en) * | 2019-05-08 | 2020-11-10 | 阿里巴巴集团控股有限公司 | Data processing method, system and device |
| CN110083688A (en) * | 2019-05-10 | 2019-08-02 | 北京百度网讯科技有限公司 | Search result recalls method, apparatus, server and storage medium |
| CN110083688B (en) * | 2019-05-10 | 2022-03-25 | 北京百度网讯科技有限公司 | Search result recall method, device, server and storage medium |
| CN110516159A (en) * | 2019-08-30 | 2019-11-29 | 北京字节跳动网络技术有限公司 | A kind of information recommendation method, device, electronic equipment and storage medium |
| CN111414540B (en) * | 2020-03-20 | 2021-01-15 | 重庆探程数字科技有限公司 | Online learning recommendation method and device, online learning system and server |
| CN111414540A (en) * | 2020-03-20 | 2020-07-14 | 张明 | Online learning recommendation method and device, online learning system and server |
| CN111444424A (en) * | 2020-03-25 | 2020-07-24 | 深圳市分期乐网络科技有限公司 | Information recommendation method and information recommendation system |
| CN113763086A (en) * | 2020-09-23 | 2021-12-07 | 北京沃东天骏信息技术有限公司 | Information recommendation method and device |
| CN112328892A (en) * | 2020-11-24 | 2021-02-05 | 北京百度网讯科技有限公司 | Information recommendation method, device, equipment and computer storage medium |
| CN112579902A (en) * | 2020-12-24 | 2021-03-30 | 第四范式（北京）技术有限公司 | Behavior data management method and device supporting multiple intelligent application scenes |
| WO2022199494A1 (en) * | 2021-03-23 | 2022-09-29 | 华为技术有限公司 | User interest-based content recommendation method, and terminal device |
| CN113094589A (en) * | 2021-04-30 | 2021-07-09 | 中国银行股份有限公司 | Intelligent service recommendation method and device |
| CN113094589B (en) * | 2021-04-30 | 2024-05-28 | 中国银行股份有限公司 | Intelligent service recommendation method and device |
| CN113190758A (en) * | 2021-05-21 | 2021-07-30 | 聚好看科技股份有限公司 | Server and media asset recommendation method |
| CN113947459A (en) * | 2021-10-21 | 2022-01-18 | 北京沃东天骏信息技术有限公司 | Article pushing and selecting method and device and storage medium |
| CN114185471A (en) * | 2022-02-17 | 2022-03-15 | 哈尔滨工业大学（威海） | Clothing recommendation method based on user intention recognition |
| CN116993412A (en) * | 2023-07-06 | 2023-11-03 | 道有道科技集团股份公司 | Intelligent delivery system and method based on user operation data analysis |
| CN116993412B (en) * | 2023-07-06 | 2024-03-01 | 道有道科技集团股份公司 | Intelligent delivery system and method based on user operation data analysis |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108763502B (en) | 2022-03-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108763502A (en) | Information recommendation method and system | |
| US11843651B2 (en) | Personalized recommendation method and system, and terminal device | |
| CN102483745B (en) | Co-selected image classification | |
| CN110543598B (en) | Information recommendation method and device and terminal | |
| CN105493057B (en) | The content selection controlled using accuracy | |
| CN109840782B (en) | Click rate prediction method, device, server and storage medium | |
| CN109889430A (en) | Information push method, device, computer equipment and storage medium | |
| CN107222566A (en) | Information-pushing method, device and server | |
| CN104899315A (en) | Method and device for pushing user information | |
| CN110413868A (en) | Information recommendation method, device, system and storage medium | |
| Lommatzsch et al. | Real-time recommendations for user-item streams | |
| CN108021673A (en) | A kind of user interest model generation method, position recommend method and computing device | |
| US20100318427A1 (en) | Enhancing database management by search, personal search, advertising, and databases analysis efficiently using core-set implementations | |
| CN109753421B (en) | Service system optimization method and device, computer equipment and storage medium | |
| CN109547807B (en) | Information processing method and device based on live broadcast and server | |
| CN110233879A (en) | Intelligently pushing interfacial process, device, computer equipment and storage medium | |
| CN110347781B (en) | Article reverse arrangement method, article recommendation method, device, equipment and storage medium | |
| CN110910201B (en) | Information recommendation control method and device, computer equipment and storage medium | |
| WO2020258773A1 (en) | Method, apparatus, and device for determining pushing user group, and storage medium | |
| CN110247974A (en) | Information-pushing method, device, computer and storage medium based on block chain | |
| CN111159563A (en) | Method, device and equipment for determining user interest point information and storage medium | |
| CN113869931A (en) | Advertisement putting strategy determining method and device, computer equipment and storage medium | |
| CN112269918A (en) | Information recommendation method, device, equipment and storage medium | |
| CN111552835A (en) | File recommendation method and device and server | |
| CN114722266A (en) | Questionnaire pushing method and device, electronic equipment and readable 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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |