CN108763502B - Information recommendation method and system - Google Patents
Information recommendation method and system Download PDFInfo
- Publication number
- CN108763502B CN108763502B CN201810538924.5A CN201810538924A CN108763502B CN 108763502 B CN108763502 B CN 108763502B CN 201810538924 A CN201810538924 A CN 201810538924A CN 108763502 B CN108763502 B CN 108763502B
- Authority
- CN
- China
- Prior art keywords
- recommendation
- real
- time
- model
- current application
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
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 an information recommendation method which comprises the steps that a terminal collects operation data of a user and sends the operation data to a server, the server obtains behavior characteristics of the user according to the operation data in a preset time length, the server constructs real-time recommendation models of application scenes corresponding to the behavior characteristics in real time and sends the real-time recommendation models to the terminal, and when a trigger condition is met, the terminal calls the real-time recommendation models matched with the characteristics of the current application scenes to obtain recommendation information and display the recommendation information. The invention also discloses an information recommendation system which can improve the recommendation accuracy and realize the recommendation based on the user individuals.
Description
Technical Field
The invention belongs to the technical field of terminals, and particularly relates to an information recommendation method and system.
Background
With the advent of the big data era, personalized recommendation algorithms based on big data are introduced successively into each big application store or application market to improve user experience. However, due to technical and cost limitations, most of these schemes utilize user portrait attributes generated by an offline mining algorithm to recommend applications to users, for example, according to the characteristics of users such as gender, age, academic history, occupation, income, and the like.
The defects of the prior art are that user portrait data are basically updated according to day-level time granularity and only can reflect long-term interest of users, and a large amount of information demands of the users have the characteristics of instant, short-time and rapid change, so that inaccurate recommendation is caused; the method is limited by the influence of factors such as user privacy protection, difficulty in collection of certain attribute data and the like, the accuracy and coverage rate of the data are difficult to improve, and inaccurate recommendation is caused; only the recommendation to a class of people can be refined, the recommendation to individuals cannot be accurate, and the personalized effect of the recommendation is poor; because the new user cannot generate historical data, the recommendation is not applicable to the new user group, and the applicability is insufficient.
Disclosure of Invention
The embodiment of the invention provides an information recommendation method and system, which can solve the problems of too long time for finding problematic application contents, low application download amount and low activity caused by manual online inspection of the application contents.
The information recommendation method provided by the embodiment of the invention comprises the following steps:
the terminal collects operation data of a user and sends the operation data to the server;
the server obtains the behavior characteristics of the user according to the operation data within a preset time length;
the server builds real-time recommendation models of all application scenes corresponding to the behavior characteristics in real time and sends the real-time recommendation models to the terminal;
and when the triggering condition is met, the terminal calls a real-time recommendation model matched with the characteristics of the current application scene to obtain recommendation information and displays the recommendation information.
An information recommendation system provided by an embodiment of the present invention includes:
a terminal and a server;
the terminal is used for acquiring operation data of a user and sending the operation data to the server;
the server is used for obtaining the behavior characteristics of the user according to the operation data within the preset duration;
the server is used for constructing a real-time recommendation model of each application scene corresponding to the behavior characteristics in real time and sending the real-time recommendation model to the terminal;
and the terminal is used for calling a real-time recommendation model matched with the characteristics of the current application scene when the triggering condition is met, obtaining recommendation information and displaying the recommendation information.
From the above embodiments of the present invention, the terminal collects the operation data of the user in real time through the client, and sends the operation data to the server, the server obtains the behavior characteristics of the user according to the operation data within the preset duration, constructs a real-time recommendation model of each application scene corresponding to the behavior characteristics and sends the real-time recommendation model to the terminal, and when a trigger condition is met, the terminal calls a real-time recommendation model matched with the characteristics of the current application scene to obtain recommendation information and display the recommendation information, the method can realize information recommendation aiming at the individual user, improve the customer experience and the information recommendation efficiency, has diversified and combinable recommendation models, is suitable for different recommendation application scenes, the method can circularly obtain the operation data of the user, continuously realize the updating of the recommendation model and realize the self-improvement of information recommendation.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below.
Fig. 1 is a schematic view of an application scenario of an information recommendation method provided in an embodiment of the present invention;
fig. 2 is a schematic flowchart of an information recommendation method according to a first embodiment of the present invention;
fig. 3 is a flowchart illustrating an information recommendation method according to a second embodiment of the present invention;
fig. 4(a), fig. 4(b), and fig. 4(c) are schematic interface diagrams of an example of an information recommendation method according to an embodiment of the present invention;
fig. 5 is a schematic structural diagram of an information recommendation system according to a third embodiment of the present invention;
fig. 6 is a schematic diagram of a terminal structure according to a fourth embodiment of the present invention.
Detailed Description
In order to make the objects, features and advantages of the present invention more obvious and understandable, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention, and it is apparent that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The information recommendation method provided by the embodiment of the invention can be applied to electronic commerce, news, reading home pages and search pages, and can also be applied to application scenes such as home pages, software pages, game pages and the like of application markets or application stores. After browsing, clicking, downloading or searching any application, commodity, news or article, the user can experience the real-time personalized recommendation result related to the previous operation through pull-down refreshing or page switching. The information recommendation method provided by the embodiment of the invention is described below by taking the operation data of the user to the application in the application market or the application store as an example.
Fig. 1 is a schematic view of a typical application scenario of an information recommendation method in each embodiment of the present invention, where a mobile terminal 10 and a server 20 are in the same network, the mobile terminal 10 acquires operation data of a user in real time through a built-in client (e.g., an application mall, an application market), and specifically acquires operation data of the user in each application scenario of the client, where the acquired operation data types include exposure data, search data, click data, download data, and the like, and the acquired operation data specifically includes: the user identifier, the operation identifier, the application identifier, the content identifier, the operation time, and the like, and specifically, the user identifier, the operation identifier, the application identifier, and the content identifier may be a user ID (identity), an operation ID, an operation object ID, and a content ID of an operation object, respectively. The collected operation data is sent to the server 20 in a single or multiple packed form for further processing.
The application scene refers to the application scenes of a home page, a software page, a game page, a search page and the like of the client.
The server 20 is a distributed server cluster, and includes an access server and a service server, where the access server is an access layer and is configured to receive operation data reported by the mobile terminal, and the access server receives the operation data reported by the mobile terminal, and then distributes the operation data to corresponding service servers according to a preset distribution rule.
The server 20 is provided with a real-time streaming computing system for real-time cleaning and computing the billions of data reported by the client. Specifically, after receiving operation data reported by a client in real time by an access layer, the real-time streaming computing system performs distributed statistics according to operation data types, that is, different types of operation data such as exposure data, search data, Click data, download data and the like are respectively counted by different servers according to different categories, and behavior characteristics of a user are counted through final summarization, wherein the behavior characteristics are indexes for evaluating behavior characteristics of the user, such as Click number, download number, real-time download conversion Rate (RCVR, real-time Click Value Rate) and the like, and the real-time download conversion Rate is a ratio of exposure to download amount in the time window. Specifically, the statistical relevant data corresponding to the user is obtained according to the user ID, the operation ID, the application ID, and the like of the user. The real-time streaming computing system stores real-time operation data of a user in a period of time recently by using a sliding time window, continuously processes the real-time operation data into lines by using a timer or a data threshold value and other triggering modes, and sends the features to the real-time model building system.
An offline model system is provided in server 20, and a plurality of offline data models are stored in the offline model system, and are generated by a conventional data mining algorithm. The offline data model comprises an application similarity model, an application correlation model, a content similarity model, a content correlation model, a user medium-long term interest model and the like, and the model refers to processing logic generated through a data mining algorithm.
The similar model of the application refers to a model of recalling the same type of application, that is, when a user operates an application, the application of the same type as the application is not recommended. If the user clicks on the first news APP, the second news APP is not recommended, and the similar model of the application is used for excluding the recommendation of the same type of application.
The related model of the application refers to a model of recalling the internally-linked applications, namely, when a user operates an application, the application associated with the application is not recommended. If the user clicks the first news APP, the first reading APP of the same operator as the first news APP is not recommended, and the correlation model of the application is used to exclude the recommendation of the associated application.
The similarity model of the content refers to a model of recalling the same type of content, that is, when the user operates on the content of an application, the content of the same type as the content is not recommended. If the user searches for a first news, a second news similar to the first news is not recommended, and a similarity model of contents is used to exclude recommendations of the same type of contents.
The correlation model of the content refers to a model of recalling the content with intrinsic relation, that is, when the user operates on the content of an application, the content associated with the content is not recommended. If the user searches for the first book, a second book of the same author as the first book is not recommended, and the correlation model of the content is used to exclude the recommendation of the associated content.
The medium-long term interest model of the user refers to a kind of medium-long term interest model of the user, and the medium-long term interest model is suitable for processing stable medium-long term operation data and is not suitable for processing real-time operation data, so that the medium-term interest model of the user is used for eliminating the recommendation of the behavior characteristics of the user.
The real-time model building system builds different real-time recommendation models according to different behavior characteristics of users and characteristics of application scenes. The application scene refers to a scene such as a home page, a search page, a software page, a game page, and the like in an application market or an application store. The feature of the application scene refers to a configuration feature of each of a home page, a search page, a software page, and a game page of an application market or an application store. The characteristics of a plurality of application scenes and each real-time recommendation model are stored in the real-time model building system.
Specifically, a plurality of recommendation models are stored in the real-time model building system, and mainly include a single-behavior model, a multi-behavior model, an exposure negative feedback model, a self-recall model, an offline model combined with the real-time model, a content-related model, a content-similar model, an application-related model, an application-similar model, a behavior-breaking model and the like.
The recommended models of the above real-time model building system are not usually used alone, but are combined to work together in a permutation and combination manner. For example, the multi-behavior model can be used for mixed sorting, and then the exposure negative feedback model is used for rearrangement, so that the advantages of the two models can be simultaneously reserved. In addition, the model can be expanded in parallel according to the data characteristics, and is not limited to the above list.
The real-time model building system builds a real-time recommendation model according to the current application scene and the real-time behavior characteristics of the current user, wherein the real-time recommendation model is recommendation logic obtained according to the current behavior characteristics and the current application scene of the user. The real-time model building system sends the built real-time recommendation model to the mobile terminal 10, the mobile terminal 10 calls the real-time recommendation model corresponding to the current scene, and the alternative data is processed according to the real-time recommendation model to obtain application recommendation information to be finally displayed to the user, which may be specifically displayed in a form of an arrangement list according to recommendations. Different real-time recommendation models can be selected according to scene characteristics in different scenes, so that the effect of maximum recommendation is achieved.
The terminal in the information recommendation method in this embodiment may include a mobile phone, a tablet computer, a PC (personal computer), and the like. For convenience of description, the terminal in each embodiment below takes a mobile terminal as an example.
Referring to fig. 2, fig. 2 is a schematic flow chart of an information recommendation method according to a first embodiment of the present invention, in this embodiment, a terminal takes a mobile phone as an example, and the method includes:
s101, a terminal collects operation data of a user and sends the operation data to a server;
specifically, the mobile phone collects data generated by operation performed by a user on a preset information display page in real time through the client, wherein the information display page is an application operation page and can be an electronic commerce page, a news page, a reading page and the like in other scenes. The operation data includes contents input by a user in the information presentation page. Such as search terms entered on a news page.
The operation data type includes any one or more of an exposure operation, a search operation, a click operation and a download operation, and the operation data specifically includes a user identifier, a behavior identifier, an operation object identifier, a content identifier of an operation object and an operation time. The user identifier, the behavior identifier, the operation object identifier, and the content identifier of the operation object may be a user ID, a behavior ID, an operation object ID, and a content ID of the operation object, where the user identifier, the behavior identifier, the operation object identifier, and the content identifier of the operation object all have unique identifiers, the operation object is, for example, an application, and the content of the operation object is, for example, application content.
The collected data is sent to a server in a single or multiple packaged form, and a real-time streaming computing system in the server can process the collected data.
S102, the server obtains the behavior characteristics of the user according to the operation data in the preset duration;
the preset time length can be determined by the server through sliding the time window, and the server acquires user operation data reported by the mobile phone within the preset time length and acquires the behavior characteristics of the user according to the operation data.
The behavior characteristics are indexes for evaluating the behavior characteristics of the user, such as click number, download number, RCVR and the like.
S103, the server constructs real-time recommendation models of all application scenes corresponding to the behavior characteristics in real time and sends the real-time recommendation models to the terminal;
various recommendation models are stored in the server, wherein the models are processing logics, and the recommendation models are logics of recommendation information.
And the server constructs a real-time recommendation model of each application scene corresponding to the behavior characteristics in real time, the real-time recommendation logic can be a combination of a plurality of recommendation models, and the constructed real-time recommendation model of each application scene is stored and sent to the terminal.
The real-time recommendation model can also be constructed by the terminal.
And S104, when the triggering condition is met, the terminal calls the real-time recommendation model matched with the characteristics of the current application scene to obtain recommendation information and displays the recommendation information.
The trigger condition may be that page turning or page refreshing is detected, the page refreshing includes an active refreshing operation of a user or passive refreshing when a preset refreshing period is reached, the page turning includes an active page turning operation of the user or passive page turning when the preset page turning period is reached, and other operations may be set as the trigger condition.
When the triggering condition is met, the mobile phone calls a real-time recommendation model matched with the characteristics of the current application scene, processes the alternative data to obtain recommendation information, wherein the recommendation information is information with a sequential arrangement order, such as a recommendation list, and displays the recommendation information according to the recommendation order.
The alternative data refers to the original data in the system corresponding to the current operation of the user, and the original data can be requested from the server by the mobile phone. For example, when the user action is to search for "WeChat" in an application marketplace, the alternative data is all WeChat-related data for that application marketplace.
In the embodiment of the invention, a terminal acquires operation data of a user in real time through a client and sends the operation data to a server, the server obtains behavior characteristics of the user according to the operation data in a preset time, the server constructs a real-time recommendation model of each application scene corresponding to the behavior characteristics and sends the real-time recommendation model to the terminal, the terminal calls the real-time recommendation model matched with the characteristics of the current application scene to obtain recommendation information and display the recommendation information, the information recommendation aiming at the user can be realized, the customer experience is improved, the information recommendation efficiency is improved, the recommendation models are diversified and combinable, the method is suitable for being applied to different recommendation application scenes, the operation data of the user can be obtained circularly, the update of the recommendation model is realized continuously, and the self-improvement of information recommendation is realized.
Referring to fig. 3, fig. 3 is a schematic flow chart of an information recommendation method according to a second embodiment of the present invention, in this embodiment, a terminal takes a mobile phone as an example, and the method includes:
s201, a terminal collects operation data of a user and sends the operation data to a server;
the operation data includes all data generated by any one or more of exposure operation, search operation, click operation and download operation, and specifically includes user identification, behavior identification, operation object identification, content identification of operation object and operation time.
S202, the server obtains the behavior characteristics of the user according to the operation data in the preset duration;
the preset time length can be determined by the server through sliding the time window, and the server acquires user operation data reported by the mobile phone within the preset time length and acquires the behavior characteristics of the user according to the operation data.
The behavior characteristics are indexes for evaluating the behavior characteristics of the user, such as click number, download number, RCVR and the like.
And performing distributed statistics on the operation data within the preset duration according to the operation data types, wherein the statistical mode is that the statistics results of the data types are summarized to obtain the behavior characteristics of the user, such as click rate, download rate, real-time download conversion rate and the like, according to the user identifier, the operation object identifier and the time window of each operation data type.
S203, acquiring recommendation models of the application scenes matched with the behavior characteristics from preset recommendation models stored in the server, taking the recommendation models of the matched application scenes as real-time recommendation models, and sending the recommendation models to the terminal;
various recommendation models are stored in the server, wherein the models are processing logics, and the recommendation models are logics of recommendation information.
The recommendation model may specifically include: the model comprises a single-line model, a multi-behavior model, an exposure negative feedback model, a self-recall model, an offline model combined with a real-time model, a content-related model, a content-similar model, an application-related model, an application-similar model, a behavior-breaking model and the like.
Wherein, the single line is a model, which refers to a processing logic for orientation only according to the most frequent behavior characteristics recently generated by the user, and the behavior characteristics become the basis for recommendation;
the multi-behavior model is processing logic for integrating a plurality of pieces of operation data recently generated by a user to perform mixed sequencing;
the exposure negative feedback model is processing logic for reducing recommendation degree of applications which are not further acted by the exposed user, for example, reducing recommendation degree of applications which are not clicked by the exposed user;
the self-recall model refers to processing logic for forcing non-recommendation of applications with behaviors of clicking, searching and the like for users;
the offline model is combined with the real-time model, and the processing logic is that the offline model in the offline model system is used for processing the alternative data of the user, and then the real-time behavior characteristics are used for processing;
the similarity model of the application refers to a model for recalling the same type of application, namely when a user operates on one application, the application of the same type as the application is not recommended. If the user clicks on the first news APP, the second news APP is not recommended, and the similar model of the application is used for excluding the recommendation of the same type of application.
The related model of the application refers to a model of recalling the internally-linked applications, namely, when a user operates an application, the application associated with the application is not recommended. If the user clicks the first news APP, the first reading APP of the same operator as the first news APP is not recommended, and the correlation model of the application is used to exclude the recommendation of the associated application.
The similarity model of the content refers to a model of recalling the same type of content, that is, when the user operates on the content of an application, the content of the same type as the content is not recommended. If the user searches for a first news, a second news similar to the first news is not recommended, and a similarity model of contents is used to exclude recommendations of the same type of contents.
The correlation model of the content refers to a model of recalling the content with intrinsic relation, that is, when the user operates on the content of an application, the content associated with the content is not recommended. If the user searches for the first book, a second book of the same author as the first book is not recommended, and the correlation model of the content is used to exclude the recommendation of the associated content.
The behavior scattering model is processing logic for disordering the operation data collected by the user between the collected real-time operation data and the collected real-time operation data according to a preset rule and then mixing the operation data and the collected real-time operation data. For example, the real-time operation data collected by the user this time is to search for a "cartoon", the operation data collected 1 day before is to search for a "secondary element", the operation data collected 1 hour before is to search for a "book purchasing website", if the behavior breaking model is applied, the user is shown with search results mixing the "cartoon", "secondary element" and "book purchasing website" when the page is refreshed, specifically, the first 3 search results showing the "cartoon", the last 3 search results showing a combination of the "cartoon" and the "secondary element", and the next group of 3 search results showing a combination of the "book purchasing website" and the "cartoon".
The above recommendation models are not typically used alone, but work in combination by permutation and combination.
The server constructs a real-time recommendation model of each application scene corresponding to the behavior characteristics, the real-time recommendation logic can be a combination of a plurality of recommendation models, and the constructed real-time recommendation model of each application scene is sent to the terminal.
Furthermore, an offline model system is also arranged in the server, and a plurality of offline data models are stored in the offline model system and are generated through a traditional data mining algorithm. The offline data model comprises an application similarity model, an application correlation model, a content similarity model, a content correlation model, a user medium-long term interest model and the like, and the model refers to processing logic generated through a data mining algorithm.
The processing logic of the similar model of the application, the related model of the application, the similar model of the content and the related model of the content in the offline model is the same as that of the similar model in the recommended model, but the different is that the similar model in the offline model is the logic for processing the offline data, and the similar model of the content, the related model of the content, the similar model of the application and the related model of the application in the real-time model building system are for processing the real-time data.
The medium-long term interest model of the user refers to a kind of medium-long term interest model of the user, and the medium-long term interest model is suitable for processing stable medium-long term operation data and is not suitable for processing real-time operation data, so that the medium-term interest model of the user is used for eliminating the recommendation of the behavior characteristics of the user. The medium and long term interest model is obtained by analyzing long term behavior characteristics of a plurality of users.
The offline data model has the following two main functions:
the method includes the steps that firstly, before calculation of real-time recommendation is carried out, alternative data are pre-screened, the alternative data refer to original data in a system corresponding to current operation of a user, the original data can be requested from a server by a mobile phone, for example, if the current operation of the user is to search for a cartoon in a client, the alternative data are all data related to the cartoon in the client. And screening a part of data from the alternative data by using an offline data model, so that the processing time delay during real-time recommendation is reduced, and the performance of a recommendation system is improved.
Secondly, under the condition that the user just opens the client without any behavior feedback, a part of data is screened from the alternative data by using the offline data model, and the personalized recommendation to the user can be realized to a certain extent.
Specifically, when a client is started for the first time, the mobile phone requests the server to load home page data, because real-time operation data are not generated yet, the server confirms an offline model as a real-time recommendation model and sends the real-time recommendation model to the mobile phone, the mobile phone calls the offline model to obtain recommendation information and display the recommendation information, and the offline model is obtained by analyzing long-term behavior characteristics of a plurality of users.
And S204, when the triggering condition is met, the terminal calls a real-time recommendation model matched with the characteristics of the current application scene to obtain recommendation information and displays the recommendation information.
The trigger condition may be that page turning or page refreshing is detected, the page refreshing includes an active refreshing operation of a user or passive refreshing when a preset refreshing period is reached, the page turning includes an active page turning operation of the user or passive page turning when the preset page turning period is reached, and other operations may be set as the trigger condition.
When the triggering condition is met, the mobile phone calls the real-time recommendation model matched with the characteristics of the current application scene, alternative data are processed, recommendation information is obtained, the recommendation information is information with a sequential arrangement order, such as a recommendation list, and the recommendation information is displayed according to the recommendation order.
The alternative data refers to the original data in the system corresponding to the current operation of the user, and the original data can be requested from the server by the mobile phone. For example, when the user operation is to search for "comics" in the client, the alternative data is all comic-related data of the client.
It should be noted that, before the mobile phone calls the real-time recommendation model to obtain the recommendation information and displays the recommendation information, the mobile phone calls a preset offline model to preprocess the alternative data, so that the processing time for obtaining the recommendation information by using the real-time recommendation model can be reduced.
The method for obtaining the recommendation information and displaying the recommendation information by calling the real-time recommendation model matched with the characteristics of the current application scene by the terminal specifically includes the steps of obtaining a target recommendation model matched with the characteristics of the current application scene, specifically obtaining the characteristics of the current application scene and obtaining operation data before page turning or page refreshing, and confirming the current application scene according to the characteristics of the current application scene, wherein the current application scene comprises a home page, a software page, a game page or a search page of a current client, and the operation data before page turning or page refreshing comprises the following steps: and browsing, clicking, downloading or searching the content in the current application scene of the user. And acquiring a target recommendation model matched with the current application scene from the stored real-time recommendation model and the offline model.
When the target recommendation model comprises the real-time recommendation model and the offline model, the mobile phone calls the offline model to preprocess the alternative data corresponding to the operation data before page turning or page refreshing, the alternative data is also the alternative data associated with the operation data before page turning or page refreshing, for example, when the current application scene is a client top page, the operation is a search, and the operation data is a cartoon, the alternative data is all the cartoon related data of the client. Further, calling a real-time recommendation model matched with the characteristics of the current application scene based on the preprocessed alternative data to obtain recommendation information and displaying the recommendation information;
and when the target recommendation model comprises a real-time recommendation model, calling the real-time recommendation model matched with the characteristics of the current application scene to obtain recommendation information and displaying the recommendation information.
Further, when a plurality of real-time recommendation models matched with the characteristics of the current application scene exist, the mobile phone calls each real-time recommendation model matched with the characteristics of the current application scene in sequence according to a preset calling sequence, and obtains recommendation information from the alternative data. Specifically, if there are 4 real-time recommendation models matching the characteristics of the current application scene, the mobile phone calls the 4 real-time recommendation models in sequence according to a preset calling sequence, and obtains final recommendation information from the alternative data.
An example is shown in fig. 4, where fig. 4(a) is a search interface for a user to search at a client using a mobile phone, fig. 4(b) is a home page returned by the user after the user searches, the home page is automatically refreshed, and fig. 4(c) is a search result showing a recommendation for the user after the refresh.
Or when the number of the real-time recommendation models matched with the characteristics of the current application scene is multiple, the terminal calls each real-time recommendation model matched with the characteristics of the current application scene respectively, multiple groups of alternative recommendation information respectively corresponding to the multiple real-time recommendation models are obtained from the alternative data, preset ranked target recommendation information in each group of the alternative recommendation information is respectively taken, the obtained multiple groups of target recommendation information are ranked according to the priority of each real-time recommendation model, and recommendation information is obtained. Specifically, if there are 4 real-time recommendation models matching the characteristics of the current application scene, the mobile phone calls the 4 real-time recommendation models respectively, obtains 4 groups of alternative recommendation information corresponding to the 4 real-time recommendation models from the alternative data, obtains preset ranked target recommendation information from the 4 groups of alternative recommendation information respectively, and ranks the obtained 4 groups of target recommendation information according to preset priorities of the real-time recommendation models to obtain final recommendation information.
Further, in order to recommend information closer to the interest of the user, the time sequence and the times of operation of each piece of recommendation information are obtained, and the priority of each real-time recommendation model is adjusted according to the time sequence and the times. The more the time sequence is, the more interested the user is in the recommendation information, the more times, the more interested the user is in the recommendation information, and the higher the priority of the real-time recommendation model is. Weights may be set for the chronological order and the degree, respectively, for example, the weights for the chronological order and the degree are 0.5 and 0.5, respectively. And adjusting the priority of each real-time recommendation model according to the set algorithm, the time sequence and the times of the operation of the user on the recommendation information and the set weight.
And the server adjusts the priority of the real-time recommendation model with the adjustment times exceeding the preset times in the server according to the adjusted priority of each real-time recommendation model sent by each terminal within the preset statistical duration. In order to eliminate accidental factors, the server can adjust the priority of the real-time recommendation model after the adjustment times exceed a certain number, and the adjustment accuracy is improved.
In the embodiment of the invention, the terminal collects the operation data of the user in real time through the client and sends the operation data to the server, the server obtains the behavior characteristics of the user according to the operation data within the preset duration, constructs a real-time recommendation model of each application scene corresponding to the behavior characteristics and sends the real-time recommendation model to the terminal, and when a trigger condition is met, the terminal calls a real-time recommendation model matched with the characteristics of the current application scene to obtain recommendation information and display the recommendation information, the method can realize information recommendation aiming at the individual user, improve the customer experience and the information recommendation efficiency, has diversified and combinable recommendation models, is suitable for different recommendation application scenes, the method can circularly obtain the operation data of the user, continuously realize the updating of the recommendation model and realize the self-improvement of information recommendation.
Referring to fig. 5, fig. 5 is a diagram illustrating an information recommendation system according to a third embodiment of the present invention, and for convenience of description, only the relevant portions of the third embodiment of the present invention are shown. The system comprises:
a terminal 301 and a server 302;
the terminal 301 is used for acquiring operation data of a user in real time and sending the operation data to the server 302;
the server 302 is configured to obtain the behavior characteristics of the user according to the operation data within a preset duration;
the server 302 is used for constructing a real-time recommendation model of each application scene corresponding to the behavior characteristics in real time and sending the real-time recommendation model to the terminal 301;
and the terminal 301 is configured to, when the trigger condition is met, call a real-time recommendation model matched with the characteristics of the current application scene to obtain recommendation information and display the recommendation information.
Further, the server 302 is further configured to determine the preset duration by sliding the time window;
the server 302 is further configured to perform distributed statistics on the operation data within the preset duration according to the operation data types, respectively;
the server 302 is further configured to aggregate the statistical result to obtain the behavior characteristic of the user.
The server 302 is further configured to send an offline model to the terminal 301 when the client is initially started, where the offline model is obtained by analyzing long-term behavior characteristics of multiple users;
the terminal 301 is further configured to invoke the offline model, obtain recommendation information, and display the recommendation information.
The terminal 301 is further configured to, when page turning or page refreshing is detected, obtain a target recommendation model matched with features of a current application scene;
the terminal 301 is further configured to obtain a feature of a current application scenario and obtain operation data before page turning or page refreshing, and determine the current application scenario according to the feature of the current application scenario, where the current application scenario includes a home page, a software page, a game page, or a search page of a current client, and the operation data before page turning or page refreshing includes: and the contents browsed, clicked, downloaded or searched in the current application scene of the user are also used for acquiring a target recommendation model matched with the current application scene.
The terminal 301 is further configured to, when the target recommendation model includes the real-time recommendation model and an offline model, invoke the offline model, pre-process the alternative data corresponding to the operation data before page turning or page refreshing of the page, and invoke the real-time recommendation model matched with the features of the current application scenario based on the pre-processed alternative data, to obtain recommendation information, and display the recommendation information;
the terminal 301 is further configured to, when the target recommendation model includes the real-time recommendation model, call the real-time recommendation model matched with the characteristics of the current application scene to obtain recommendation information and display the recommendation information.
The terminal 301 is further configured to, when a plurality of real-time recommendation models matched with the features of the current application scene are provided, sequentially call, according to a preset call sequence, each real-time recommendation model matched with the features of the current application scene, and obtain the recommendation information from the alternative data.
The terminal 301 is further configured to, when a plurality of real-time recommendation models matched with the features of the current application scene are available, respectively invoke each real-time recommendation model matched with the features of the current application scene, and obtain a plurality of sets of alternative recommendation information respectively corresponding to the plurality of real-time recommendation models from the alternative data;
and respectively taking preset ranked target recommendation information in each group of the alternative recommendation information, and ranking the obtained multiple groups of target recommendation information according to the priority of each real-time recommendation model to obtain recommendation information.
The terminal 301 is further configured to obtain a time sequence and times of operations of each piece of recommendation information, adjust priorities of each real-time recommendation model according to the time sequence and times, and send the adjusted priorities of each real-time recommendation model to the server 302;
the server 302 is further configured to adjust, in the server 302, the priority of the real-time recommendation model whose adjustment times exceed the preset times according to the adjusted priority of each real-time recommendation model sent by each terminal 301 within the preset statistical duration.
The undescribed technical details in the embodiments of the present invention are the same as those in the embodiments shown in fig. 1 to 4, and are not described again here.
In the embodiment of the invention, the terminal collects the operation data of the user in real time through the client and sends the operation data to the server, the server obtains the behavior characteristics of the user according to the operation data within the preset duration, constructs a real-time recommendation model of each application scene corresponding to the behavior characteristics and sends the real-time recommendation model to the terminal, and when a trigger condition is met, the terminal calls a real-time recommendation model matched with the characteristics of the current application scene to obtain recommendation information and display the recommendation information, the method can realize information recommendation aiming at the individual user, improve the customer experience and the information recommendation efficiency, has diversified and combinable recommendation models, is suitable for different recommendation application scenes, the method can circularly obtain the operation data of the user, continuously realize the updating of the recommendation model and realize the self-improvement of information recommendation.
Referring to fig. 6, fig. 6 is a terminal according to a fourth embodiment of the present invention, and for convenience of description, only the portions related to the embodiment of the present invention are shown. The terminal includes:
an acquisition module 401, configured to acquire operation data of a user;
a sending module 402, configured to send the operation data to a server;
and the recommending module 403 is configured to receive the real-time recommending model of each application scene sent by the server, and when a triggering condition is met, call the real-time recommending model matched with the characteristics of the current application scene to obtain recommending information and display the recommending information.
Further, the recommending module 403 is further configured to, when the client is initially started, obtain an offline model sent by the server, where the offline model is obtained by analyzing long-term behavior characteristics of multiple users, and call the offline model to obtain recommendation information and display the recommendation information.
A further advancement of the terminal includes:
the obtaining module 404 is further configured to obtain a target recommendation model matched with the features of the current application scenario when page turning or page refreshing is detected;
the obtaining module 404 is further configured to obtain a feature of a current application scenario and obtain operation data before page turning or page refreshing, and determine the current application scenario according to the feature of the current application scenario, where the current application scenario includes a home page, a software page, a game page, or a search page of a current client, and the operation data before page turning or page refreshing includes: browsing, clicking, downloading or searching contents in the current application scene of a user; and acquiring a target recommendation model matched with the current application scene.
The recommending module 403 is further configured to, when the target recommending model includes the real-time recommending model and an offline model, invoke the offline model, pre-process the candidate data corresponding to the operation data before page turning or page refreshing of the page, and invoke, based on the pre-processed candidate data, the real-time recommending model matched with the features of the current application scenario, to obtain recommending information and display the recommending information;
the recommending module 403 is further configured to, when the target recommending model includes the real-time recommending model, call a real-time recommending model matched with characteristics of a current application scene to obtain recommending information and display the recommending information;
the recommending module 403 is further configured to, when there are multiple real-time recommending models matching the features of the current application scene, sequentially invoke, according to a preset invoking sequence, each real-time recommending model matching the features of the current application scene, and obtain the recommending information from the alternative data.
The recommending module 403 is further configured to invoke a preset offline model to preprocess the alternative data.
The recommending module 403 is further configured to, when there are multiple real-time recommending models matching the features of the current application scene, respectively invoke each real-time recommending model matching the features of the current application scene by the terminal, and obtain multiple sets of alternative recommending information respectively corresponding to the multiple real-time recommending models from the alternative data; and respectively taking preset ranked target recommendation information in each group of the alternative recommendation information, and ranking the obtained multiple groups of target recommendation information according to the priority of each real-time recommendation model to obtain recommendation information.
The obtaining module 404 is further configured to obtain a time sequence and a number of times that each piece of recommendation information is operated;
an adjusting module 405, configured to adjust the priority of each real-time recommendation model according to the time sequence and the number of times;
the sending module 402 is further configured to send the adjusted priorities of the real-time recommendation models to the server.
The embodiments of the present invention have not been described in detail, but refer to the same description.
In the embodiment of the invention, the terminal collects the operation data of the user in real time through the client and sends the operation data to the server, so that the server obtains the behavior characteristics of the user according to the operation data in the preset duration, and constructing a real-time recommendation model of each application scene corresponding to the behavior characteristics, receiving the real-time recommendation model sent by the server by the terminal, and when the triggering condition is met, the terminal calls a real-time recommendation model matched with the characteristics of the current application scene to obtain recommendation information and display the recommendation information, the method can realize information recommendation aiming at the individual user, improve the customer experience and the information recommendation efficiency, has diversified and combinable recommendation models, is suitable for different recommendation application scenes, the method can circularly obtain the operation data of the user, continuously realize the updating of the recommendation model and realize the self-improvement of information recommendation.
The embodiment of the present application further provides a computer-readable storage medium, where the computer-readable storage medium may be provided in the electronic device in each of the above embodiments. The computer-readable storage medium has stored thereon a computer program which, when executed by a processor, implements the information recommendation method described in the foregoing embodiments shown in fig. 2 and 3. Further, the computer-readable storage medium may be various media that can store program codes, such as a usb disk, a removable hard disk, a Read-Only Memory (ROM), a RAM, a magnetic disk, or an optical disk.
In the above embodiments, the descriptions of the respective embodiments have respective emphasis, and for parts that are not described in detail in a certain embodiment, reference may be made to related descriptions of other embodiments.
In the above description, for a person skilled in the art, there are variations on the specific implementation and application scope according to the ideas of the embodiments of the present invention, and in summary, the content of the present specification should not be construed as a limitation to the present invention.
Claims (9)
1. An information recommendation method, comprising:
the terminal collects operation data of a user in real time and sends the operation data to the server;
the server obtains the behavior characteristics of the user according to the operation data within a preset time length;
the server builds real-time recommendation models of all application scenes corresponding to the behavior characteristics in real time and sends the real-time recommendation models to the terminal;
when a triggering condition is met, the terminal calls a real-time recommendation model matched with the characteristics of the current application scene to obtain recommendation information and display the recommendation information, wherein the triggering condition is that page turning or page refreshing is detected and comprises the following steps:
when page turning or page refreshing is detected, acquiring a target recommendation model matched with the characteristics of the current application scene;
when the target recommendation model comprises the real-time recommendation model and an offline model, the terminal calls the offline model, screens alternative data corresponding to operation data before page turning or page refreshing, calls a real-time recommendation model matched with the characteristics of a current application scene based on the screened alternative data, obtains recommendation information and displays the recommendation information;
and when the target recommendation model comprises the real-time recommendation model, calling the real-time recommendation model matched with the characteristics of the current application scene to obtain recommendation information and displaying the recommendation information.
2. The method of claim 1, wherein the obtaining, by the server, the behavior characteristic of the user according to the operation data within a preset time period comprises:
the server determines the preset duration through a sliding time window;
performing distributed statistics on the operation data within the preset duration according to the operation data types respectively;
and summarizing the statistical result to obtain the behavior characteristics of the user.
3. The method of claim 1, wherein the obtaining a target recommendation model that matches features of a current application scenario comprises:
acquiring the characteristics of a current application scene and acquiring operation data before page turning or page refreshing of the page, and confirming the current application scene according to the characteristics of the current application scene, wherein the current application scene comprises a home page, a software page, a game page or a search page of a current client, and the operation data before page turning or page refreshing comprises: browsing, clicking, downloading or searching the content in the current application scene by the user;
and acquiring a target recommendation model matched with the current application scene.
4. The method of claim 1, wherein the method further comprises:
when a client is started for the first time, the server sends an offline model to the terminal, and the offline model is obtained by analyzing long-term behavior characteristics of a plurality of users;
and the terminal calls the offline model to obtain recommendation information and displays the recommendation information.
5. The method of claim 1, wherein the terminal calls a real-time recommendation model matched with the characteristics of the current application scene to obtain recommendation information comprises:
and when a plurality of real-time recommendation models matched with the characteristics of the current application scene exist, the terminal sequentially calls each real-time recommendation model matched with the characteristics of the current application scene according to a preset calling sequence, and acquires the recommendation information from the alternative data.
6. The method of claim 1, wherein the terminal calls a real-time recommendation model matched with the characteristics of the current application scene to obtain recommendation information comprises:
when a plurality of real-time recommendation models matched with the characteristics of the current application scene exist, the terminal calls each real-time recommendation model matched with the characteristics of the current application scene respectively, and a plurality of groups of alternative recommendation information respectively corresponding to the real-time recommendation models are obtained from alternative data;
and respectively taking preset ranked target recommendation information in each group of the alternative recommendation information, and ranking the obtained multiple groups of target recommendation information according to the priority of each real-time recommendation model to obtain the recommendation information.
7. The method of claim 6, wherein the terminal invokes a real-time recommendation model matching features of a current application scenario, and after obtaining recommendation information, the method further comprises:
acquiring the time sequence and the times of operation of each piece of recommendation information;
adjusting the priority of each real-time recommendation model according to the time sequence and the times, and sending the adjusted priority of each real-time recommendation model to the server;
and the server adjusts the priority of the real-time recommendation model with the adjustment times exceeding the preset times in the server according to the adjusted priority of each real-time recommendation model sent by each terminal within the preset statistical duration.
8. An information recommendation system, comprising:
a terminal and a server;
the terminal is used for acquiring operation data of a user in real time and sending the operation data to the server;
the server is used for obtaining the behavior characteristics of the user according to the operation data within the preset duration;
the server is used for constructing a real-time recommendation model of each application scene corresponding to the behavior characteristics in real time and sending the real-time recommendation model to the terminal;
the terminal is used for calling a real-time recommendation model matched with the characteristics of the current application scene when a trigger condition is met, obtaining recommendation information and displaying the recommendation information, wherein the trigger condition is that page turning or page refreshing is detected and comprises the following steps:
when page turning or page refreshing is detected, acquiring a target recommendation model matched with the characteristics of the current application scene;
when the target recommendation model comprises the real-time recommendation model and an offline model, calling the offline model, screening alternative data corresponding to operation data before page turning or page refreshing, calling the real-time recommendation model matched with the characteristics of the current application scene based on the screened alternative data, obtaining recommendation information and displaying the recommendation information;
and when the target recommendation model comprises the real-time recommendation model, calling the real-time recommendation model matched with the characteristics of the current application scene to obtain recommendation information and displaying the recommendation information.
9. The system of claim 8,
the server is further used for determining the preset duration through a sliding time window;
the server is further configured to perform distributed statistics on the operation data within the preset duration according to the operation data types;
the server is also used for summarizing the statistical result to obtain the behavior characteristics of the user;
the server is further used for sending an offline model to the terminal when the client is started for the first time, wherein the offline model is obtained by analyzing long-term behavior characteristics of a plurality of users;
the terminal is also used for calling the offline model to obtain recommendation information and displaying the recommendation information;
the terminal is further configured to acquire a feature of a current application scene, acquire the operation data before page turning or page refreshing, and confirm the current application scene according to the feature of the current application scene, where the current application scene includes a home page, a software page, a game page, or a search page of a current client, and the operation data before page turning or page refreshing includes: browsing, clicking, downloading or searching the content in the current application scene by the user;
and acquiring a target recommendation model matched 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 CN108763502A (en) | 2018-11-06 |
| CN108763502B true 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) |
Families Citing this family (19)
| 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 |
| 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 |
| 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 |
| 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 |
| CN110516159B (en) * | 2019-08-30 | 2022-12-20 | 北京字节跳动网络技术有限公司 | Information recommendation method and device, electronic equipment and storage medium |
| CN112287228A (en) * | 2020-03-20 | 2021-01-29 | 张明 | Online learning recommendation method, 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 |
| CN112328892B (en) * | 2020-11-24 | 2024-06-18 | 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 |
| CN115114515A (en) * | 2021-03-23 | 2022-09-27 | 华为技术有限公司 | Content recommendation method based on user interest and terminal equipment |
| CN113094589B (en) * | 2021-04-30 | 2024-05-28 | 中国银行股份有限公司 | Intelligent service recommendation method and device |
| CN113190758B (en) * | 2021-05-21 | 2023-01-20 | 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 |
| 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 |
Citations (2)
| 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 |
| 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 |
Family Cites Families (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7440943B2 (en) * | 2000-12-22 | 2008-10-21 | Xerox Corporation | Recommender system and method |
| CN104123284B (en) * | 2013-04-24 | 2018-01-23 | 华为技术有限公司 | The method and server of a kind of recommendation |
| CN103886090B (en) * | 2014-03-31 | 2018-01-02 | 北京搜狗科技发展有限公司 | Content recommendation method and device based on user preferences |
| CN104021483B (en) * | 2014-06-26 | 2017-08-25 | 陈思恩 | Passenger demand recommends method |
| CN105589905B (en) * | 2014-12-26 | 2019-06-18 | ***股份有限公司 | The analysis of user interest data and collection system and its method |
| CN105183781B (en) * | 2015-08-14 | 2018-11-20 | 百度在线网络技术（北京）有限公司 | Information recommendation method and device |
| CN107423442B (en) * | 2017-08-07 | 2020-09-25 | 火烈鸟网络(广州)股份有限公司 | Application recommendation method and system based on user portrait behavior analysis, storage medium and computer equipment |
| 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 (2)
| 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 |
| 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 |
Non-Patent Citations (1)
| Title |
|---|
| 基于分布式流处理框架下的移动健身管理***研究;孙凯涛;《现代电子技术》;20161101(第21期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108763502A (en) | 2018-11-06 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108763502B (en) | Information recommendation method and system | |
| CN107222566B (en) | Information pushing method and device and server | |
| WO2020048084A1 (en) | Resource recommendation method and apparatus, computer device, and computer-readable storage medium | |
| US11843651B2 (en) | Personalized recommendation method and system, and terminal device | |
| WO2017071251A1 (en) | Information pushing method and device | |
| EP3873065B1 (en) | Content recommendation method, mobile terminal, and server | |
| US20140279121A1 (en) | Customizable and adjustable pricing of games | |
| CN110413867B (en) | Method and system for content recommendation | |
| CN107808314B (en) | User recommendation method and device | |
| CN110543598A (en) | information recommendation method and device and terminal | |
| CN106980703A (en) | For the method and device of group's search, electronic equipment, computer-readable medium | |
| JP6154963B2 (en) | Information processing apparatus, information processing method, and information processing program | |
| CN111105269A (en) | Advertisement putting processing method, device, equipment and storage medium | |
| CN112104505B (en) | Application recommendation method, device, server and computer readable storage medium | |
| CN111523035B (en) | Recommendation method, device, server and medium for APP browsing content | |
| CN111159553A (en) | Information pushing method and device, computer equipment and storage medium | |
| CN111552835B (en) | File recommendation method, device and server | |
| CN110858377A (en) | Information processing method, page display method, system and equipment | |
| CN112463994A (en) | Multimedia resource display method, device, system and storage medium | |
| CN116955817A (en) | Content recommendation method, device, electronic equipment and storage medium | |
| CN113326436B (en) | Method, device, electronic equipment and storage medium for determining recommended resources | |
| CN112000865B (en) | Hotword generation method, device, server and storage medium | |
| CN109271580B (en) | Search method, device, client and search engine | |
| CN113850416A (en) | Advertisement promotion cooperation object determining method and device | |
| CN106484710B (en) | Dynamic data processing method and device and information display method and device |
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 |