CN109408724B - Method and device for determining pre-estimated click rate of multimedia resource and server - Google Patents

Method and device for determining pre-estimated click rate of multimedia resource and server Download PDF

Info

Publication number
CN109408724B
CN109408724B CN201811314076.6A CN201811314076A CN109408724B CN 109408724 B CN109408724 B CN 109408724B CN 201811314076 A CN201811314076 A CN 201811314076A CN 109408724 B CN109408724 B CN 109408724B
Authority
CN
China
Prior art keywords
information
click rate
multimedia
click
multimedia resource
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
Application number
CN201811314076.6A
Other languages
Chinese (zh)
Other versions
CN109408724A (en
Inventor
牛亚男
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Beijing Dajia Internet Information Technology Co Ltd
Original Assignee
Beijing Dajia Internet Information Technology Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Beijing Dajia Internet Information Technology Co Ltd filed Critical Beijing Dajia Internet Information Technology Co Ltd
Priority to CN201811314076.6A priority Critical patent/CN109408724B/en
Publication of CN109408724A publication Critical patent/CN109408724A/en
Priority to PCT/CN2019/105452 priority patent/WO2020093781A1/en
Application granted granted Critical
Publication of CN109408724B publication Critical patent/CN109408724B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q30/00—Commerce
    • G06Q30/02—Marketing; Price estimation or determination; Fundraising
    • G06Q30/0201—Market modelling; Market analysis; Collecting market data
    • G06Q30/0202—Market predictions or forecasting for commercial activities
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • General Physics & Mathematics (AREA)
  • Strategic Management (AREA)
  • Finance (AREA)
  • Development Economics (AREA)
  • Accounting & Taxation (AREA)
  • Data Mining & Analysis (AREA)
  • Computational Linguistics (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • General Health & Medical Sciences (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Evolutionary Computation (AREA)
  • Biophysics (AREA)
  • Biomedical Technology (AREA)
  • Artificial Intelligence (AREA)
  • Molecular Biology (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Game Theory and Decision Science (AREA)
  • Economics (AREA)
  • Marketing (AREA)
  • General Business, Economics & Management (AREA)
  • Information Transfer Between Computers (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)
  • User Interface Of Digital Computer (AREA)

Abstract

The disclosure relates to a method, a device and a server for determining an estimated click rate of a multimedia resource, and belongs to the field of information recommendation. The method comprises the following steps: acquiring user behavior information of a current user; acquiring multimedia attribute information of a first multimedia resource, wherein the first multimedia resource is a multimedia resource to be recommended to the current user; calling a click rate prediction model, wherein the click rate prediction model comprises an embedded layer and a click rate prediction network, the embedded layer comprises a weight matrix corresponding to at least one information type, and the click rate prediction network is used for outputting the predicted click rate of the multimedia resource by taking an embedded vector output by the embedded layer as input; inputting the user behavior information and the multimedia attribute information into the click rate estimation model, and outputting the estimated click rate of the user on the first multimedia resource. By adopting the method and the device, the accuracy of the estimated click rate can be improved.

Description

Method and device for determining pre-estimated click rate of multimedia resource and server
Technical Field
The disclosure relates to the field of information recommendation, in particular to a method, a device and a server for determining an estimated click rate of a multimedia resource.
Background
In an information recommendation system, the related technology can predict the Click Through Rate (CTR) of multimedia resources and display the multimedia resources with higher predicted Click Through Rate to a user, so that the probability of the user clicking the multimedia resources is improved, and the accuracy of information recommendation is improved.
When the click rate of the multimedia resource is estimated, a click rate estimation model based on a Deep Neural Network (DNN) is commonly used in the related art to calculate the estimated click rate. The input data of the click through rate prediction model can be divided into different fields, such as the field of multimedia resources and the field of users. As shown in the schematic diagram of the click-through rate prediction model shown in fig. 1, input data may be converted into low-dimensional embedded vectors through an embedding (embedding) layer, where different domains may correspond to different weight matrices in the embedding layer, and each item of feature information in one domain may determine the embedded vectors through the corresponding weight matrices. The feature information and the embedded vector have a one-to-one correspondence relationship, and therefore, the weight matrix corresponding to a domain may also be referred to as an embedding mapping table of the domain. Finally, the embedded vector can be input into a deep neural network in the related technology, and the estimated click rate of the multimedia resource can be output.
However, in a domain, the number of some feature information may be smaller, for example, when the feature information in the user domain includes the picture identifier clicked by the user, if the number of times of clicking the picture by the user is smaller, the number of the captured picture identifiers is also smaller. When the click rate estimation model is trained, the learning of the corresponding weight matrix is insufficient due to the small number of characteristic information, that is, the representativeness of the obtained embedded vector is weak, so that the accuracy of the obtained estimated click rate is low when the click rate of the multimedia resource is estimated.
Disclosure of Invention
The invention provides a method, a device and a server for determining an estimated click rate of a multimedia resource, which can solve the problem of low accuracy of the estimated click rate.
According to a first aspect of the embodiments of the present disclosure, a method for determining an estimated click rate of a multimedia resource is provided, including:
acquiring user behavior information of a user;
acquiring multimedia attribute information of a first multimedia resource, wherein the first multimedia resource is a multimedia resource to be recommended to the user;
calling a click rate prediction model, wherein the click rate prediction model comprises an embedded layer and a click rate prediction network, the embedded layer comprises a weight matrix corresponding to at least one information type, and the click rate prediction network is used for outputting the predicted click rate of the multimedia resource by taking an embedded vector output by the embedded layer as input;
inputting the user behavior information and the multimedia attribute information into the click rate pre-estimation network, and outputting the pre-estimated click rate of the user to the first multimedia resource.
Optionally, the inputting the user behavior information and the multimedia attribute information into the click rate estimation model and outputting the estimated click rate of the user on the first multimedia resource includes:
for each information type, inputting information belonging to the information type in the user behavior information and the multimedia attribute information into a weight matrix corresponding to the information type in the embedding layer, and outputting at least one embedding vector;
and inputting at least one embedding vector output by the embedding layer into the click rate prediction model, and outputting the predicted click rate of the user on the first multimedia resource.
Optionally, the training method of the click rate prediction model includes:
acquiring an initial model of the click rate estimation model;
acquiring at least one training sample, wherein the training sample comprises multimedia attribute information of a second multimedia resource, user behavior information when a sample user browses the second multimedia resource, and a click condition of the sample user on the second multimedia resource, and the click condition comprises clicked or un-clicked;
and training the initial model based on the at least one training sample to obtain the click rate estimation model.
Optionally, the initial model includes an initial embedding layer and an initial click rate pre-estimation network, where the initial embedding layer includes an initial weight matrix corresponding to at least one information type;
the training the initial model based on the at least one training sample to obtain the click rate estimation model includes:
for an initial weight matrix corresponding to each information type, performing parameter adjustment on the initial weight matrix based on a training sample containing the information type to obtain a trained weight matrix corresponding to the information type;
adjusting parameters of the initial click rate pre-estimation network based on the at least one training sample to obtain a trained click rate pre-estimation network;
and obtaining the click rate estimation model based on the weight matrix corresponding to the trained at least one information type and the trained click rate estimation network.
Optionally, in the parameter adjustment process of the initial weight matrix corresponding to each information type, when the number of training samples including a first information type is smaller than the number of training samples including a second information type, the learning rate corresponding to the first information type is greater than the learning rate corresponding to the second information type.
Optionally, the information type includes a work identifier, an author identifier, and/or a style identifier.
Optionally, the user behavior information includes click history information, attention information and/or favorite information, the click history information is used for representing multimedia attribute information of a multimedia resource clicked by a user, the attention information is used for representing multimedia attribute information of a multimedia resource concerned by the user, and the favorite information is used for representing multimedia attribute information of a multimedia resource favorite by the user.
According to a second aspect of the embodiments of the present disclosure, there is provided an apparatus for determining an estimated click rate of a multimedia resource, including:
an acquisition unit configured to acquire user behavior information of a user;
the acquiring unit is further configured to acquire multimedia attribute information of a first multimedia resource, where the first multimedia resource is a multimedia resource to be recommended to the user;
the click rate prediction network is used for taking an embedded vector output by the embedded layer as input and outputting the predicted click rate of the multimedia resource;
and the determining unit is configured to input the user behavior information and the multimedia attribute information into the click rate estimation model and output the estimated click rate of the user on the first multimedia resource.
Optionally, the determining unit is configured to:
for each information type, inputting information belonging to the information type in the user behavior information and the multimedia attribute information into a weight matrix corresponding to the information type in the embedding layer, and outputting at least one embedding vector;
and inputting at least one embedding vector output by the embedding layer into the click rate pre-estimated network, and outputting the pre-estimated click rate of the user on the first multimedia resource.
Optionally, the apparatus further comprises a training unit configured to:
acquiring an initial model of the click rate estimation model;
acquiring at least one training sample, wherein the training sample comprises multimedia attribute information of a second multimedia resource, user behavior information when a sample user browses the second multimedia resource, and a click condition of the sample user on the second multimedia resource, and the click condition comprises clicked or un-clicked;
and training the initial model based on the at least one training sample to obtain the click rate estimation model.
Optionally, the initial model includes an initial embedding layer and an initial click rate pre-estimation network, where the initial embedding layer includes an initial weight matrix corresponding to at least one information type;
the training unit is configured to:
for an initial weight matrix corresponding to each information type, performing parameter adjustment on the initial weight matrix based on a training sample containing the information type to obtain a trained weight matrix corresponding to the information type;
adjusting parameters of the initial click rate pre-estimation network based on the at least one training sample to obtain a trained click rate pre-estimation network;
and obtaining the click rate estimation model based on the weight matrix corresponding to the trained at least one information type and the trained click rate estimation network.
Optionally, in the parameter adjustment process of the initial weight matrix corresponding to each information type, when the number of training samples including a first information type is smaller than the number of training samples including a second information type, the learning rate corresponding to the first information type is greater than the learning rate corresponding to the second information type.
Optionally, the information type includes a work identifier, an author identifier, and/or a style identifier.
Optionally, the user behavior information includes click history information, attention information and/or favorite information, the click history information is used for representing multimedia attribute information of a multimedia resource clicked by a user, the attention information is used for representing multimedia attribute information of a multimedia resource concerned by the user, and the favorite information is used for representing multimedia attribute information of a multimedia resource favorite by the user.
According to a third aspect of the embodiments of the present disclosure, there is provided a server, including:
one or more processors;
one or more memories for storing one or more processor-executable instructions;
wherein the one or more processors are configured to:
acquiring user behavior information of a user;
acquiring multimedia attribute information of a first multimedia resource, wherein the first multimedia resource is a multimedia resource to be recommended to the user;
calling a click rate prediction model, wherein the click rate prediction model comprises an embedded layer and a click rate prediction network, the embedded layer comprises a weight matrix corresponding to at least one information type, and the click rate prediction network is used for outputting the predicted click rate of the multimedia resource by taking an embedded vector output by the embedded layer as input;
inputting the user behavior information and the multimedia attribute information into the click rate estimation model, and outputting the estimated click rate of the user on the first multimedia resource.
According to a fourth aspect of the embodiments of the present disclosure, there is provided a non-transitory computer-readable storage medium, wherein instructions, when executed by a processor of a server, enable the server to perform a method for determining a pre-estimated click rate of a multimedia resource, the method comprising:
acquiring user behavior information of a user;
acquiring multimedia attribute information of a first multimedia resource, wherein the first multimedia resource is a multimedia resource to be recommended to the user;
calling a click rate prediction model, wherein the click rate prediction model comprises an embedded layer and a click rate prediction network, the embedded layer comprises a weight matrix corresponding to at least one information type, and the click rate prediction network is used for outputting the predicted click rate of the multimedia resource by taking an embedded vector output by the embedded layer as input;
inputting the user behavior information and the multimedia attribute information into the click rate estimation model, and outputting the estimated click rate of the user on the first multimedia resource.
According to a fifth aspect of the embodiments of the present disclosure, there is provided an application program/computer program product, when the application program/computer program product is running on a server, causing the server to execute a method for determining an estimated click rate of a multimedia resource, the method comprising:
acquiring user behavior information of a current user;
acquiring multimedia attribute information of a first multimedia resource, wherein the first multimedia resource is a multimedia resource to be recommended to the current user;
calling a click rate prediction model, wherein the click rate prediction model comprises an embedded layer and a click rate prediction network, the embedded layer comprises a weight matrix corresponding to at least one information type, and the click rate prediction network is used for outputting the predicted click rate of the multimedia resource by taking an embedded vector output by the embedded layer as input;
inputting the user behavior information and the multimedia attribute information into the click rate estimation model, and outputting the estimated click rate of the user on the first multimedia resource.
The technical scheme provided by the embodiment of the disclosure can have the following beneficial effects:
in the user behavior information and the multimedia attribute information, the information of the same information type can determine the embedded vector through the same weight matrix in the embedded layer, and the representativeness of the embedded vector can be improved. Therefore, when the click rate of the multimedia resource is estimated based on the method of the embodiment, the accuracy of the estimated click rate is improved.
It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the disclosure.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present disclosure and together with the description, serve to explain the principles of the disclosure.
FIG. 1 is a diagram illustrating a click through rate prediction model according to an exemplary embodiment.
FIG. 2 is a diagram illustrating an implementation environment in accordance with an example embodiment.
FIG. 3 is a flowchart illustrating a method for determining an estimated click rate of a multimedia asset according to an exemplary embodiment.
FIG. 4 is a flowchart illustrating a method for determining an estimated click rate of a multimedia asset according to an exemplary embodiment.
FIG. 5 is a schematic diagram illustrating an application interface in accordance with an illustrative embodiment.
FIG. 6 is a diagram illustrating a click-through rate prediction model according to an exemplary embodiment.
FIG. 7 is a diagram illustrating a click through rate prediction model according to an exemplary embodiment.
FIG. 8 is a flowchart illustrating a method for training a click rate prediction model, according to an example embodiment.
FIG. 9 is a flowchart illustrating a method for training a click rate prediction model according to an exemplary embodiment.
FIG. 10 is a block diagram illustrating an apparatus for determining a pre-estimated click rate of a multimedia resource according to an exemplary embodiment.
FIG. 11 is a block diagram illustrating an apparatus for determining an estimated click rate of a multimedia asset according to an example embodiment.
Detailed Description
Reference will now be made in detail to the exemplary embodiments, examples of which are illustrated in the accompanying drawings. When the following description refers to the accompanying drawings, like numbers in different drawings represent the same or similar elements unless otherwise indicated. The implementations described in the exemplary embodiments below are not intended to represent all implementations consistent with the present disclosure. Rather, they are merely examples of apparatus and methods consistent with certain aspects of the present disclosure, as detailed in the appended claims.
The embodiment provides an implementation environment diagram of a method for determining an estimated click rate of a multimedia resource, and the implementation environment diagram is shown in fig. 2. The implementation environment may include a plurality of terminals 201, a server 202 for providing services to the plurality of terminals 201. A plurality of terminals 201 are connected to the server 202 through a wireless or wired network, and the plurality of terminals 201 may be computer devices or intelligent terminals or the like capable of accessing the server 202. The terminal 201 may have an application program installed therein for recommending multimedia resources (such as pictures, short videos, etc.), and a user may log in the application program. The server 202 may provide background services for the above applications and record user behavior information of each user. The server 202 may further have at least one database for storing click through rate estimation models, multimedia resources and corresponding multimedia attribute information, user behavior information of each user, and the like.
The embodiment provides a method for determining an estimated click rate of a multimedia resource, which may be implemented by a server, and as shown in a flowchart of the method for determining an estimated click rate of a multimedia resource shown in fig. 3, a processing flow of the method may include the following steps:
in step S301, the server acquires user behavior information of the user.
In step S302, the server acquires multimedia attribute information of a first multimedia asset.
The first multimedia resource is a multimedia resource to be recommended to the user.
In step S303, the server invokes a click through rate estimation model.
The click rate prediction model comprises an embedding layer and a click rate prediction network, wherein the embedding layer comprises at least one weight matrix corresponding to one information type, and the click rate prediction network is used for outputting the predicted click rate of the multimedia resource by taking an embedded vector output by the embedding layer as input.
In step S304, the server inputs the user behavior information and the multimedia attribute information into the click-through rate estimation model, and outputs the estimated click-through rate of the user on the first multimedia resource.
Optionally, the inputting the user behavior information and the multimedia attribute information into the click rate estimation network, and outputting the estimated click rate of the user to the first multimedia resource includes:
for each information type, inputting user behavior information and information belonging to the information type in the multimedia attribute information into a weight matrix corresponding to the information type in the embedded layer, and outputting at least one embedded vector;
inputting at least one embedding vector output by the embedding layer into a click rate estimation model, and outputting the estimated click rate of the user on the first multimedia resource.
Optionally, the training method of the click rate prediction model includes:
acquiring an initial model of a click rate estimation model;
acquiring at least one training sample, wherein the training sample comprises multimedia attribute information of a second multimedia resource, user behavior information when a sample user browses the second multimedia resource and click conditions of the sample user on the second multimedia resource, and the click conditions comprise clicked or un-clicked conditions;
and training the initial model based on at least one training sample to obtain a click rate estimation model.
Optionally, the initial model includes an initial embedding layer and an initial click rate pre-estimation network, where the initial embedding layer includes an initial weight matrix corresponding to at least one information type;
the training of the initial model based on at least one training sample to obtain the click rate estimation model comprises the following steps:
for the initial weight matrix corresponding to each information type, performing parameter adjustment on the initial weight matrix based on a training sample containing the information type to obtain a weight matrix corresponding to the trained information type;
adjusting parameters of the initial click rate pre-estimation network based on at least one training sample to obtain a trained click rate pre-estimation network;
and obtaining a click rate estimation model based on the weight matrix corresponding to the trained at least one information type and the trained click rate estimation network.
Optionally, in the parameter adjustment process of the initial weight matrix corresponding to each information type, when the number of training samples including the first information type is smaller than the number of training samples including the second information type, the learning rate corresponding to the first information type is greater than the learning rate corresponding to the second information type.
Optionally, the information type includes a composition identification, an author identification, and/or a style identification.
Optionally, the user behavior information includes click history information, attention information and/or favorite information, the click history information includes information recorded when the user clicks any multimedia resource, the attention information includes information recorded when the user clicks an attention option, and the favorite information includes information recorded when the user clicks a favorite option.
The embodiment will be described with reference to a specific embodiment, with reference to a method for determining an estimated click rate of a multimedia resource. The method may be implemented by a server, for example, as shown in fig. 4, a flowchart of a method for determining an estimated click rate of a multimedia resource, where a processing flow of the method may include the following steps:
in step S401, the server acquires user behavior information of the user.
When a user starts an application program, a terminal can send a multimedia resource acquisition request to a server; or, when the user searches for the multimedia resource through the application program, the terminal may also send an acquisition request of the multimedia resource to the server. When receiving an acquisition request of a multimedia resource sent by a terminal, a server may trigger a processing logic for recommending the multimedia resource, so as to perform processing by the determination method for estimating the click rate of the multimedia resource provided in this embodiment. The embodiment does not limit the specific manner of triggering the processing logic of recommending multimedia resources.
At this time, the server may obtain the user behavior information of the user from the stored user behavior information of each user according to the identification information of the user. In one possible implementation, the user behavior information may include click history information, attention information and/or preference information, the click history information may be used to represent multimedia attribute information of a multimedia resource clicked by the user, the attention information may be used to represent multimedia attribute information of a multimedia resource concerned by the user, and the preference information may be used to represent multimedia attribute information of a multimedia resource preferred by the user.
Optionally, the information type may include a composition identifier, an author identifier and/or a style identifier for multimedia attribute information of the multimedia asset recorded by the server.
The following describes the process of the server recording the above information:
as shown in the application interface diagram of fig. 5, the terminal may present a multimedia asset provided by the application, and may include a focus option and a favorite option in a presentation interface of the multimedia asset.
When the user clicks and views the multimedia resource, the server can receive a request for loading the multimedia resource, and further can add corresponding multimedia attribute information to click history information of the user for storage. For example, when a user clicks to view a short video, the server may add a work identification, author identification, and/or genre identification for the short video to the user's click history information.
If the user needs to subscribe to the multimedia resource, the user can click on the attention option in the presentation interface. Furthermore, the server can receive the attention adding request of the multimedia resource, and add the corresponding multimedia attribute information to the attention information of the user for storage. Of course, the attention option may be not only the attention for the multimedia resource, but also the attention for the author or style of the multimedia resource, which is not limited in this embodiment. For example, if the attention option is attention directed to the author of the multimedia asset, the server may add the author identification of the multimedia asset to the attention information of the user, after which, when the author updates the work, the user may receive a corresponding update notification, achieving the effect of subscription.
If the user likes the multimedia resource, the user can click the favorite option in the display interface, and the processing of adding the corresponding favorite information by the server is the same as the processing of adding the attention information, which is not described again.
Of course, the above-mentioned schemes of recording the click history information, the attention information and the favorite information are all alternatives, that is, the user behavior information recorded by the server may be one or more of the click history information, the attention information and the favorite information. The information type contained in the information may also be one or more of a work identifier, an author identifier, and a style identifier, and may also include other information types, and the specific information type is not limited in this embodiment.
In step S402, the server acquires multimedia attribute information of a first multimedia asset.
The first multimedia resource is a multimedia resource to be recommended to the user. The multimedia attribute information may be obtained by the author through active addition when the author uploads the corresponding multimedia resource, or may be automatically generated by the server according to the author, the content and other information of the multimedia resource.
The first multimedia asset may comprise a plurality of multimedia assets, for example, may be currently popular multimedia assets, or may be search results from a search for the user. Because the multimedia resources which can be viewed by the user are limited, the server can determine which first multimedia resources are displayed and can also sequence the displayed first multimedia resources so as to achieve the purpose of preferentially displaying the multimedia resources which are more in line with the requirements of the user. In this embodiment, the estimated click rate of the multimedia resource is used to achieve the above object, and the higher the estimated click rate is, the higher the possibility that the user clicks the multimedia resource is, that is, the multimedia resource meets the requirements of the user.
When determining the estimated click rate of the first multimedia resource, for a first multimedia resource, the server may obtain the multimedia attribute information of the first multimedia resource from the stored multimedia resource and the corresponding multimedia attribute information according to the identification information of the first multimedia resource.
In step S403, the server invokes a click through rate prediction model.
As shown in the schematic diagram of the click-through rate estimation model shown in fig. 6, the click-through rate estimation model provided in this embodiment may include an embedding layer and a click-through rate estimation network, the embedding layer may include a weight matrix corresponding to at least one information type, and the click-through rate estimation network may be configured to output the estimated click-through rate of the multimedia resource by using an embedded vector output by the embedding layer as an input. The click prediction network may be a deep neural network or a convolutional neural network, and a technician may design the click rate prediction network according to a requirement, and the specific network structure is not limited in this embodiment.
Of course, a plurality of click rate estimation models may also be stored in the server, and the server may invoke a click rate estimation model satisfying a preset condition, for example, when the server further stores the accuracy of each click rate estimation model, any click rate estimation model satisfying the accuracy greater than a preset threshold may be invoked. The present embodiment does not limit the specific manner of invoking the click rate model.
The click rate estimation model can be trained periodically, and the latest click rate estimation model obtained through training can be stored in the server. When the server needs to determine the estimated click rate of the first multimedia resource, the latest click rate estimation model can be called.
In step S404, for each information type, the server inputs information belonging to the information type in the user behavior information and the multimedia attribute information into a weight matrix corresponding to the information type in the embedding layer, and outputs at least one embedding vector.
The server may input the user behavior information acquired in step S401 and the multimedia attribute information of the first multimedia resource acquired in step S402 into the embedding layer of the click rate estimation model.
In the embedding layer, the input data may be encoded to accommodate the computation of the neural network. Generally, the dimensionality of the vector obtained after encoding is large, so that the vector after encoding can be subjected to dimensionality reduction through the weight matrix in the embedded layer, so as to reduce IO (Input-Output) overhead in the processing process. For example, a feature information of the input data is subjected to one-hot encoding to obtain a vector (0,0,0,1,0,0, 0), and the vector is converted into an embedded vector (0.145,0.152) through a weight matrix.
Each information type has a corresponding weight matrix, and in the user behavior information and the multimedia attribute information, the server can acquire vectors corresponding to the feature information of the same information type, input the vectors corresponding to the same information type into the corresponding weight matrix, and output the embedded vectors.
As already introduced in step S401 above, in one possible implementation, the information type may include a composition identification, an author identification, and/or a genre identification. For example, as shown in the schematic diagram of the click-through rate estimation model shown in fig. 7, the user behavior information is click history information, each feature information is a work identifier clicked by the user, and the feature information of the multimedia attribute information is a work identifier of the multimedia resource. The embedding mapping table shown in fig. 7 is a weight matrix corresponding to the work identifier, and the server may determine the embedded vector by using both the work identifier in the user behavior information and the work identifier in the multimedia attribute information through the embedding mapping table. That is, the weight matrices in the embedded layers used by the information of the product identifier are the same and cannot be distinguished by the domain of the multimedia resource or the domain of the user.
In step S405, the server inputs at least one embedded vector output by the embedding layer into the click-through rate prediction network, and outputs the predicted click-through rate of the user for the first multimedia resource.
After the server determines each embedded vector of the input data, the embedded vector can be input into the click rate prediction network, data processing is carried out through each network node in the click rate prediction network, and the predicted click rate of the first multimedia resource is output.
For each first multimedia resource, the estimated click rate may be determined through the above steps S402-S405. The plurality of first multimedia resources may determine the estimated click rate in a parallel processing manner, and the order of determining the estimated click rate by the plurality of first multimedia resources is not limited in this embodiment.
After determining the estimated click rate of each first multimedia resource, the server may rank each first multimedia resource according to a sequence of the estimated click rates from large to small, and may send each first multimedia resource and the corresponding rank to the terminal. Furthermore, the terminal may display the received first multimedia resource according to the ordering. Optionally, when the number of the displayed multimedia resources is the preset number, the server may send the first multimedia resources with the preset number ranked in the top and the corresponding ranking to the terminal, and the terminal may display the first multimedia resources with the preset number according to the ranking.
Of course, the first multimedia resource sent by the server to the terminal may be in a thumbnail form corresponding to the first multimedia resource, so as to reduce consumption of network resources, for example, when the first multimedia resource is a picture, the thumbnail of the picture may be sent by the server to the terminal; when the first multimedia resource is a short video, the server sends a preview to the terminal. The specific form of the first multimedia resource transmitted in this embodiment is not limited.
After the terminal displays the first multimedia resource, the current user preferentially sees the first multimedia resource with higher estimated click rate. Therefore, the determination method for the estimated click rate of the multimedia resource provided by the embodiment can improve the click rate of the user on the recommended multimedia resource. When the method is applied to the application program, the user retention rate of the application program can be improved.
In this embodiment, in the user behavior information and the multimedia attribute information, information of the same information type may determine the embedded vector through the same weight matrix in the embedded layer, and the representativeness of the embedded vector may be improved. Therefore, when the click rate of the multimedia resource is estimated based on the method of the embodiment, the accuracy of the estimated click rate is improved.
In the above embodiment, a process of determining the estimated click rate by using the click rate estimation model is described, and before using the click rate estimation model, the click rate estimation model may be trained. The embodiment provides a training method of a click rate estimation model, which can be implemented by a server. As shown in fig. 8, a flowchart of a training method of a click rate estimation model, a processing flow of the method may include the following steps:
in step S801, the server obtains an initial model of the click-through rate estimation model.
The initial model may include an initial embedding layer and an initial click-through rate prediction network, where the initial embedding layer includes an initial weight matrix corresponding to at least one information type, corresponding to the click-through rate prediction model described in the above embodiment.
The server may store an initial model of the click-through rate prediction model. The initial model can be a machine learning model designed by technicians and used for determining the estimated click rate, user behavior information and multimedia attribute information are used as input, the click rate of the user on multimedia resources is predicted, and the estimated click rate is output. However, since the model parameters in the initial model are all preset initial values, the accuracy of the predicted click rate is low, and the initial model needs to be trained.
In step S802, the server obtains at least one training sample.
The training sample may include multimedia attribute information of the second multimedia resource, user behavior information of the sample user when browsing the second multimedia resource, and a click condition of the sample user on the second multimedia resource, where the click condition may include a clicked condition or an un-clicked condition. That is, the second multimedia asset may refer to a multimedia asset that is historically presented to the sample user.
The following describes the process of the server recording information in the training sample:
the multimedia attribute information of the second multimedia asset can be recorded whenever the server transmits the second multimedia asset for presentation to the terminal. Of course, the server may also record the identification information of the second multimedia resource, and the identification information may be used to obtain the multimedia attribute information of the second multimedia resource.
At this time, the server may further obtain user behavior information of the user, and may record the user behavior information in correspondence with the information (such as multimedia attribute information of the second multimedia asset, or identification information of the second multimedia asset) recorded by the server.
When the user clicks and views any second multimedia resource, the server may receive a request for loading the second multimedia resource, and further may record the click condition of the second multimedia resource as clicked, and record corresponding to the information, for example, record corresponding to the multimedia attribute information, the user behavior information, and the click condition of the second multimedia resource.
When the terminal closes the display interface, a display closing notification can be sent to the server, and when the server receives the display closing notification, the server can acquire the second non-clicked multimedia resource from the sent second multimedia resources and record the corresponding click condition as non-click. Or, the server may further obtain the second multimedia resource that is not clicked when the preset time length is reached after the second multimedia resource for display is sent to the terminal. The embodiment does not limit the specific manner in which the server obtains the second multimedia resource that is not clicked.
When the server trains the click rate estimation model, the multimedia attribute information of the second multimedia resource, the user behavior information when the user browses the second multimedia resource and the click condition of the user on the second multimedia resource, which are recorded in the process, can be obtained and used as training samples. Optionally, the server may use the training sample whose click condition is clicked as a positive sample, and use the training sample whose click condition is not clicked as a negative sample.
In step S803, the server trains the initial model based on at least one training sample to obtain a click rate estimation model.
For each training sample, the server can input multimedia attribute information and user behavior information in the training sample into an initial model, and perform data processing based on model parameters of each network node in the initial model to obtain an estimated click rate of the initial model to a second multimedia resource. Then, the server can determine the gradient of each model parameter in the initial model according to the click condition of the user on the second multimedia resource in the training sample and the corresponding estimated click rate. The server can determine the correction value of each model parameter according to the gradient of each model parameter, and carry out parameter adjustment on each model parameter based on the correction value, namely error back propagation.
In a possible embodiment, as shown in the flowchart of the training method of the click rate estimation model in fig. 9, step S803 may include steps S8031-S8033:
in step S8031, for the initial weight matrix corresponding to each information type, the server performs parameter adjustment on the initial weight matrix based on the training sample including the information type, to obtain a weight matrix corresponding to the trained information type.
The individual model parameters in the weight matrix may be adjusted during the training process. In the above embodiment, it has been described that the feature information of the same information type may determine the embedding vector based on the same weight matrix, and further determine the estimated click rate. Correspondingly, in the training process, for the initial weight matrix corresponding to each information type, the server can obtain the training sample for determining the estimated click rate through the initial weight matrix and the corresponding estimated click rate, further calculate the gradient of each model parameter in the initial weight matrix according to the estimated click rate and the actual click condition of the user, determine the corresponding correction value according to the gradient, and adjust each model parameter. After training is finished, the obtained weight matrix corresponding to each information type can be used for determining the embedded vector for the corresponding information.
Compared with the scheme of dividing the weight matrix based on the field of the multimedia resource and the field of the user, because the information of the same information type uses the same weight matrix, the weight matrix can be trained by fully utilizing the training sample containing the information type, so that the weight matrix can be fully learned.
In step S8032, the server adjusts parameters of the initial click rate estimation network based on at least one training sample, so as to obtain a trained click rate estimation network.
Generally, each training sample can perform data processing through the click rate prediction network, so that the server can obtain each training sample and the corresponding predicted click rate, and perform parameter adjustment on the initial click rate prediction network, and the specific processing is as described above, and is not described herein again.
In step S8033, the server obtains a click rate estimation model based on the trained weight matrix corresponding to the at least one information type and the trained click rate estimation network.
When the condition of finishing training is reached (for example, the preset training times or the value of the loss function is smaller than the target value), the server can acquire each weight matrix in the current embedded layer, the click rate estimation network and the like to form a click rate estimation model, and the click rate estimation model can be stored. When the server needs to predict the multimedia resources, the stored click rate prediction model can be obtained for processing.
Of course, after that, the server may also train the stored click rate estimation model again, and the training process is the same as the above process. The server continuously updates the click rate estimation model, so that the accuracy of the click rate estimation model can be improved.
In a possible implementation manner, in the parameter adjustment process of the initial weight matrix corresponding to each information type, when the number of training samples including the first information type is smaller than the number of training samples including the second information type, the learning rate corresponding to the first information type is greater than the learning rate corresponding to the second information type.
In step S8031, when determining the correction value of each model parameter in the initial weight matrix according to the gradient, the server may adjust the learning rate according to the gradient, for example, the learning rate adjustment method may be an adaptive learning rate (adaptive learning rate) algorithm.
Generally, if the training samples containing the information type are fewer, the determined gradient is smaller, that is, the gradient changes more smoothly, and the server may increase the learning rate corresponding to the information type, that is, increase the amplitude of the correction value, so that the weight matrix is updated with a more sufficient gradient. By the method, model parameters can be fully learned under the condition of sparse characteristics, and the accuracy of the model is improved.
Of course, the method for adjusting the learning rate may also be applied to the step S8032, so that the model parameter in the click rate estimation network may also be adjusted according to the gradient adaptability, and the method for adjusting the parameter of the click rate estimation network is not limited in this embodiment.
In one possible embodiment, the training goal of the server on the initial model may be to maximize the AUC (area under the ROC Curve; ROC, Receiver Operating characteristics).
If a training sample with a click case of clicked is referred to as a first training sample and a training sample with a click case of unchecked is referred to as a second training sample, the AUC may refer to the probability that the first training sample is ranked before the second training sample.
After the server determines the estimated click rate for each training sample through the initial model, the estimated click rate can be arranged in a descending order, and the value of the AUC can be further determined according to the number of the first training samples arranged before all the second training samples and the total number of the training samples. The greater the AUC, the more the first training samples are ranked before all the second training samples, i.e. the higher the accuracy of the click rate estimation model.
Of course, the server may also determine the AUC in other manners, for example, after an ROC curve is established based on the estimated click rate of the training sample, an area under the ROC curve is calculated by using an integration method. The present embodiment is not limited to the specific manner of determining AUC.
Experiments show that the method provided by the embodiment can obviously improve the AUC, namely the click rate estimation model accuracy obtained by the method provided by the embodiment is improved.
In this embodiment, the weight matrix corresponding to one information type may be trained based on a training sample including the information type. Because the characteristic information of the same information type is not divided based on the field of multimedia resources and the field of users, training samples can be fully utilized, so that the weight matrix of the embedded layer is fully learned, the representativeness of the embedded vector is improved, and the accuracy of the click rate estimation model is improved.
FIG. 10 is a block diagram illustrating an apparatus for determining a pre-estimated click rate of a multimedia resource according to an exemplary embodiment. Referring to fig. 10, the apparatus includes an acquisition unit 1010, a calling unit 1020, and a determination unit 1030.
The acquiring unit 1010 configured to acquire user behavior information of a user;
the obtaining unit 1010 is further configured to obtain multimedia attribute information of a first multimedia resource, where the first multimedia resource is a multimedia resource to be recommended to the user;
the invoking unit 1020 is configured to invoke a click-through rate pre-estimation model, where the click-through rate pre-estimation model includes an embedded layer and a click-through rate pre-estimation network, the embedded layer includes a weight matrix corresponding to at least one information type, and the click-through rate pre-estimation network is configured to output a pre-estimated click-through rate of a multimedia resource by using an embedded vector output by the embedded layer as an input;
the determining unit 1030 is configured to input the user behavior information and the multimedia attribute information into the click rate prediction model, and output the predicted click rate of the user on the first multimedia resource.
Optionally, the determining unit 1030 is configured to:
for each information type, inputting information belonging to the information type in the user behavior information and the multimedia attribute information into a weight matrix corresponding to the information type in the embedding layer, and outputting at least one embedding vector;
and inputting at least one embedding vector output by the embedding layer into the click rate prediction model, and outputting the predicted click rate of the user on the first multimedia resource.
Optionally, the apparatus further comprises a training unit configured to:
acquiring an initial model of the click rate estimation model;
acquiring at least one training sample, wherein the training sample comprises multimedia attribute information of a second multimedia resource, user behavior information when a user browses the second multimedia resource, and a click condition of the user on the second multimedia resource, and the click condition comprises clicked or un-clicked;
and training the initial model based on the at least one training sample to obtain the click rate estimation model.
Optionally, the initial model includes an initial embedding layer and an initial click rate pre-estimation network, where the initial embedding layer includes an initial weight matrix corresponding to at least one information type;
the training unit is configured to:
for an initial weight matrix corresponding to each information type, performing parameter adjustment on the initial weight matrix based on a training sample containing the information type to obtain a trained weight matrix corresponding to the information type;
adjusting parameters of the initial click rate pre-estimation network based on the at least one training sample to obtain a trained click rate pre-estimation network;
and obtaining the click rate estimation model based on the weight matrix corresponding to the trained at least one information type and the trained click rate estimation network.
Optionally, in the parameter adjustment process of the initial weight matrix corresponding to each information type, when the number of training samples including a first information type is smaller than the number of training samples including a second information type, the learning rate corresponding to the first information type is greater than the learning rate corresponding to the second information type.
Optionally, the information type includes a work identifier, an author identifier, and/or a style identifier.
Optionally, the user behavior information includes click history information, attention information and/or favorite information, the click history information is used for representing multimedia attribute information of a multimedia resource clicked by a user, the attention information is used for representing multimedia attribute information of a multimedia resource concerned by the user, and the favorite information is used for representing multimedia attribute information of a multimedia resource favorite by the user.
With regard to the apparatus in the above-described embodiment, the specific manner in which each module performs the operation has been described in detail in the embodiment related to the method, and will not be elaborated here.
In this embodiment, in the user behavior information and the multimedia attribute information, information of the same information type may determine the embedded vector through the same weight matrix in the embedded layer, and the representativeness of the embedded vector may be improved. Therefore, when the click rate of the multimedia resource is estimated based on the method of the embodiment, the accuracy of the click rate estimation model is improved.
FIG. 11 is a block diagram illustrating an apparatus 1100 for determining an estimated click rate of a multimedia asset according to an example embodiment. For example, the apparatus 1100 may be provided as a server. Referring to fig. 11, the apparatus 1100 includes a processing component 1122 that further includes one or more processors and memory resources, represented by memory 1132, for storing instructions, such as application programs, executable by the processing component 1122. The application programs stored in memory 1132 may include one or more modules that each correspond to a set of instructions. In addition, the processing component 1122 is configured to execute the instructions to perform the method for determining the estimated click rate of the multimedia resource.
The apparatus 1100 may also include a power component 1126 configured to perform power management of the apparatus 1100, a wired or wireless network interface 1150 configured to connect the apparatus 1100 to a network, and an input/output (I/O) interface 1158. The apparatus 1100 may operate based on an operating system stored in the memory 1132, such as Windows Server, MacOS XTM, UnixTM, LinuxTM, FreeBSDTM, or the like.
In an exemplary embodiment, a non-transitory computer readable storage medium, such as a memory, including instructions executable by a processor in a server to perform the method for determining an estimated click rate of a multimedia resource is also provided. For example, the computer readable storage medium may be a ROM, a Random Access Memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, an optical data storage device, and the like.
In an exemplary embodiment, an application/computer program product is also provided that includes one or more instructions executable by a processor of a server to perform the method for determining an estimated click rate of a multimedia resource described above.
Other embodiments of the disclosure will be apparent to those skilled in the art from consideration of the specification and practice of the disclosure disclosed herein. This application is intended to cover any variations, uses, or adaptations of the disclosure following, in general, the principles of the disclosure and including such departures from the present disclosure as come within known or customary practice within the art to which the disclosure pertains. It is intended that the specification and examples be considered as exemplary only, with a true scope and spirit of the disclosure being indicated by the following claims.
It will be understood that the present disclosure is not limited to the precise arrangements described above and shown in the drawings and that various modifications and changes may be made without departing from the scope thereof. The scope of the present disclosure is limited only by the appended claims.

Claims (12)

1. A method for determining an estimated click rate of a multimedia resource is characterized by comprising the following steps:
acquiring user behavior information of a user;
acquiring multimedia attribute information of a first multimedia resource, wherein the first multimedia resource is a multimedia resource to be recommended to the user;
calling a click rate estimation model, wherein the click rate estimation model comprises an embedded layer and a click rate estimation network, the embedded layer comprises a weight matrix corresponding to at least one information type, the information type comprises a work identifier, an author identifier and/or a style identifier, and the click rate estimation network is used for outputting the estimated click rate of the multimedia resource by taking an embedded vector output by the embedded layer as input;
for each information type, inputting information belonging to the information type in the user behavior information and the multimedia attribute information into a weight matrix corresponding to the information type in the embedding layer, and outputting at least one embedding vector;
and inputting at least one embedding vector output by the embedding layer into the click rate pre-estimated network, and outputting the pre-estimated click rate of the user on the first multimedia resource.
2. The method of claim 1, wherein the training method of the click-through rate estimation model comprises:
acquiring an initial model of the click rate estimation model;
acquiring at least one training sample, wherein the training sample comprises multimedia attribute information of a second multimedia resource, user behavior information when a sample user browses the second multimedia resource, and a click condition of the sample user on the second multimedia resource, and the click condition comprises clicked or un-clicked;
and training the initial model based on the at least one training sample to obtain the click rate estimation model.
3. The method of claim 2, wherein the initial model comprises an initial embedding layer and an initial click-through rate prediction network, the initial embedding layer comprises an initial weight matrix corresponding to at least one information type;
the training the initial model based on the at least one training sample to obtain the click rate estimation model includes:
for an initial weight matrix corresponding to each information type, performing parameter adjustment on the initial weight matrix based on a training sample containing the information type to obtain a trained weight matrix corresponding to the information type;
adjusting parameters of the initial click rate pre-estimation network based on the at least one training sample to obtain a trained click rate pre-estimation network;
and obtaining the click rate estimation model based on the weight matrix corresponding to the trained at least one information type and the trained click rate estimation network.
4. The method according to claim 3, wherein in the parameter adjustment process of the initial weight matrix corresponding to each information type, when the number of training samples containing a first information type is smaller than the number of training samples containing a second information type, the learning rate corresponding to the first information type is greater than the learning rate corresponding to the second information type.
5. The method according to any one of claims 1 to 4, wherein the user behavior information includes click history information, attention information and/or preference information, the click history information is used for representing multimedia attribute information of a multimedia resource clicked by the user, the attention information is used for representing multimedia attribute information of a multimedia resource concerned by the user, and the preference information is used for representing multimedia attribute information of a multimedia resource liked by the user.
6. An apparatus for determining an estimated click rate of a multimedia resource, comprising:
an acquisition unit configured to acquire user behavior information of a user;
the acquiring unit is further configured to acquire multimedia attribute information of a first multimedia resource, where the first multimedia resource is a multimedia resource to be recommended to the user;
the click rate prediction network is used for taking an embedded vector output by the embedded layer as input and outputting the predicted click rate of the multimedia resource;
the determining unit is configured to input information belonging to the information type in the user behavior information and the multimedia attribute information into a weight matrix corresponding to the information type in the embedding layer and output at least one embedding vector for each information type; and inputting at least one embedding vector output by the embedding layer into the click rate pre-estimated network, and outputting the pre-estimated click rate of the user on the first multimedia resource.
7. The apparatus of claim 6, further comprising a training unit configured to:
acquiring an initial model of the click rate estimation model;
acquiring at least one training sample, wherein the training sample comprises multimedia attribute information of a second multimedia resource, user behavior information when a sample user browses the second multimedia resource, and a click condition of the sample user on the second multimedia resource, and the click condition comprises clicked or un-clicked;
and training the initial model based on the at least one training sample to obtain the click rate estimation model.
8. The apparatus of claim 7, wherein the initial model comprises an initial embedding layer and an initial click-through rate prediction network, the initial embedding layer comprises an initial weight matrix corresponding to at least one information type;
the training unit is configured to:
for an initial weight matrix corresponding to each information type, performing parameter adjustment on the initial weight matrix based on a training sample containing the information type to obtain a trained weight matrix corresponding to the information type;
adjusting parameters of the initial click rate pre-estimation network based on the at least one training sample to obtain a trained click rate pre-estimation network;
and obtaining the click rate estimation model based on the weight matrix corresponding to the trained at least one information type and the trained click rate estimation network.
9. The apparatus according to claim 8, wherein in the parameter adjustment process of the initial weight matrix corresponding to each information type, when the number of training samples containing a first information type is smaller than the number of training samples containing a second information type, the learning rate corresponding to the first information type is greater than the learning rate corresponding to the second information type.
10. The apparatus according to any one of claims 6 to 9, wherein the user behavior information includes click history information, attention information and/or preference information, the click history information is used to indicate multimedia attribute information of a multimedia asset clicked by the user, the attention information is used to indicate multimedia attribute information of a multimedia asset concerned by the user, and the preference information is used to indicate multimedia attribute information of a multimedia asset liked by the user.
11. A server, comprising:
one or more processors;
one or more memories for storing one or more processor-executable instructions;
wherein the one or more processors are configured to execute the instructions to implement the method for determining the estimated multimedia resource click rate of any one of claims 1-5.
12. A non-transitory computer readable storage medium, wherein instructions in the storage medium, when executed by a processor of a server, enable the server to perform a method of determining estimated multimedia asset click rates as claimed in any one of claims 1 to 5.
CN201811314076.6A 2018-11-06 2018-11-06 Method and device for determining pre-estimated click rate of multimedia resource and server Active CN109408724B (en)

Priority Applications (2)

Application Number Priority Date Filing Date Title
CN201811314076.6A CN109408724B (en) 2018-11-06 2018-11-06 Method and device for determining pre-estimated click rate of multimedia resource and server
PCT/CN2019/105452 WO2020093781A1 (en) 2018-11-06 2019-09-11 Multimedia resource estimated click through rate determination method and apparatus, and server

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201811314076.6A CN109408724B (en) 2018-11-06 2018-11-06 Method and device for determining pre-estimated click rate of multimedia resource and server

Publications (2)

Publication Number Publication Date
CN109408724A CN109408724A (en) 2019-03-01
CN109408724B true CN109408724B (en) 2020-10-02

Family

ID=65471674

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201811314076.6A Active CN109408724B (en) 2018-11-06 2018-11-06 Method and device for determining pre-estimated click rate of multimedia resource and server

Country Status (2)

Country Link
CN (1) CN109408724B (en)
WO (1) WO2020093781A1 (en)

Families Citing this family (31)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109408724B (en) * 2018-11-06 2020-10-02 北京达佳互联信息技术有限公司 Method and device for determining pre-estimated click rate of multimedia resource and server
CN110297970B (en) * 2019-05-28 2021-04-23 北京达佳互联信息技术有限公司 Information recommendation model training method and device
CN112150182B (en) * 2019-06-28 2023-08-29 腾讯科技（深圳）有限公司 Multimedia file pushing method and device, storage medium and electronic device
CN110472071A (en) * 2019-07-03 2019-11-19 中移（杭州）信息技术有限公司 Multimedia file recommendation method, device, the network equipment and storage medium
CN112307313B (en) * 2019-07-31 2024-04-16 北京达佳互联信息技术有限公司 Information recommendation method, device and storage medium
CN110502715B (en) * 2019-08-28 2023-07-14 腾讯科技（深圳）有限公司 Click probability prediction method and device
CN110825956A (en) * 2019-09-17 2020-02-21 中国平安人寿保险股份有限公司 Information flow recommendation method and device, computer equipment and storage medium
CN110909182B (en) * 2019-11-29 2023-05-09 北京达佳互联信息技术有限公司 Multimedia resource searching method, device, computer equipment and storage medium
CN111191165A (en) * 2019-12-05 2020-05-22 腾讯科技（深圳）有限公司 Data processing method, device, equipment and computer readable storage medium
CN112925924A (en) * 2019-12-05 2021-06-08 北京达佳互联信息技术有限公司 Multimedia file recommendation method and device, electronic equipment and storage medium
CN111160959B (en) * 2019-12-20 2023-06-27 深圳前海微众银行股份有限公司 User click conversion prediction method and device
CN111340522B (en) * 2019-12-30 2024-03-08 支付宝实验室(新加坡)有限公司 Resource recommendation method, device, server and storage medium
CN113495966B (en) * 2020-03-18 2023-06-23 北京达佳互联信息技术有限公司 Interactive operation information determining method and device and video recommendation system
CN113742567B (en) * 2020-05-29 2023-08-22 北京达佳互联信息技术有限公司 Recommendation method and device for multimedia resources, electronic equipment and storage medium
CN111708944A (en) * 2020-06-17 2020-09-25 北京达佳互联信息技术有限公司 Multimedia resource identification method, device, equipment and storage medium
CN111737573A (en) * 2020-06-17 2020-10-02 北京三快在线科技有限公司 Resource recommendation method, device, equipment and storage medium
CN113836391B (en) * 2020-06-24 2023-07-11 北京达佳互联信息技术有限公司 Resource recommendation method, device, computer equipment and storage medium
CN114077701A (en) * 2020-08-13 2022-02-22 北京达佳互联信息技术有限公司 Method and device for determining resource information, computer equipment and storage medium
CN112364184B (en) * 2020-11-12 2024-04-30 北京达佳互联信息技术有限公司 Method, device, server and storage medium for ordering multimedia data
CN112256918B (en) * 2020-11-17 2021-07-06 中国计量大学 Short video click rate prediction method based on multi-mode dynamic routing
CN112395504B (en) * 2020-12-01 2021-11-23 中国计量大学 Short video click rate prediction method based on sequence capsule network
CN112732953B (en) * 2020-12-30 2024-04-12 上海众源网络有限公司 Recommendation method, sample analysis device, electronic equipment and storage medium
CN112839247B (en) * 2021-01-04 2023-01-20 聚好看科技股份有限公司 Content display method, display device and cloud server
CN112950266B (en) * 2021-02-26 2024-05-03 北京达佳互联信息技术有限公司 Media resource display method and device
CN113254679B (en) * 2021-06-28 2021-09-28 北京达佳互联信息技术有限公司 Multimedia resource recommendation method and device, electronic equipment and storage medium
CN113722588B (en) * 2021-08-12 2023-09-05 北京达佳互联信息技术有限公司 Resource recommendation method and device and electronic equipment
CN113516522B (en) * 2021-09-14 2022-03-04 腾讯科技（深圳）有限公司 Media resource recommendation method, and training method and device of multi-target fusion model
CN114547417A (en) * 2022-02-25 2022-05-27 北京百度网讯科技有限公司 Media resource ordering method and electronic equipment
CN114548076A (en) * 2022-02-28 2022-05-27 深圳市和讯华谷信息技术有限公司 Intelligent scoring method for content file and related device
CN114677168A (en) * 2022-03-11 2022-06-28 百果园技术(新加坡)有限公司 Resource recommendation method, device, equipment and medium
CN114780843A (en) * 2022-04-20 2022-07-22 北京字节跳动网络技术有限公司 Model training method, medium resource recommendation method, device, medium and equipment

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104462594A (en) * 2014-12-29 2015-03-25 北京奇虎科技有限公司 Method and device for providing user personalized resource message pushing
CN106604068A (en) * 2016-12-30 2017-04-26 中广热点云科技有限公司 Method and system for updating media program

Family Cites Families (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102663617A (en) * 2012-03-20 2012-09-12 亿赞普(北京)科技有限公司 Method and system for prediction of advertisement clicking rate
CN104951965B (en) * 2015-06-26 2017-04-19 深圳市腾讯计算机系统有限公司 Advertisement delivery method and device
CN105183772A (en) * 2015-08-07 2015-12-23 百度在线网络技术（北京）有限公司 Release information click rate estimation method and apparatus
CN105095964B (en) * 2015-08-17 2017-10-20 杭州朗和科技有限公司 A kind of data processing method and device
CN105678335B (en) * 2016-01-08 2019-07-02 车智互联（北京）科技有限公司 It estimates the method, apparatus of clicking rate and calculates equipment
CN107463580B (en) * 2016-06-06 2020-07-24 腾讯科技（深圳）有限公司 Click rate estimation model training method and device and click rate estimation method and device
CN105975641A (en) * 2016-07-15 2016-09-28 合网络技术（北京）有限公司 Video recommendation method ad device
CN108304441B (en) * 2017-11-14 2022-04-26 腾讯科技（深圳）有限公司 Network resource recommendation method and device, electronic equipment, server and storage medium
CN108307207A (en) * 2017-12-28 2018-07-20 北京达佳互联信息技术有限公司 A kind of video pushing method and device
CN109408724B (en) * 2018-11-06 2020-10-02 北京达佳互联信息技术有限公司 Method and device for determining pre-estimated click rate of multimedia resource and server

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104462594A (en) * 2014-12-29 2015-03-25 北京奇虎科技有限公司 Method and device for providing user personalized resource message pushing
CN106604068A (en) * 2016-12-30 2017-04-26 中广热点云科技有限公司 Method and system for updating media program

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
从FM推演各深度CTR预估模型(附代码);龙心尘、寒小阳;《https://www.jiqizhixin.com/articles/2018-07-16-17》;20180716;第1-35页 *

Also Published As

Publication number Publication date
CN109408724A (en) 2019-03-01
WO2020093781A1 (en) 2020-05-14

Similar Documents

Publication Publication Date Title
CN109408724B (en) Method and device for determining pre-estimated click rate of multimedia resource and server
US11727048B2 (en) Methods, systems, and media for presenting recommended content based on social cues
US9838743B2 (en) Techniques for context aware video recommendation
CN111105819B (en) Clipping template recommendation method and device, electronic equipment and storage medium
CN109086439B (en) Information recommendation method and device
US8839303B2 (en) System and method for enhancing user search results by determining a television program currently being displayed in proximity to an electronic device
US20180020250A1 (en) Recommendation information pushing method, server, and storage medium
US10380649B2 (en) System and method for logistic matrix factorization of implicit feedback data, and application to media environments
CN108595493B (en) Media content pushing method and device, storage medium and electronic device
US20160171036A1 (en) Item recommendation method and apparatus
CN109168047B (en) Video recommendation method and device, server and storage medium
KR102141674B1 (en) Optimization of content distribution using model
CN111858969B (en) Multimedia data recommendation method, device, computer equipment and storage medium
US20230004608A1 (en) Method for content recommendation and device
CN113641835B (en) Multimedia resource recommendation method and device, electronic equipment and medium
CN113254679B (en) Multimedia resource recommendation method and device, electronic equipment and storage medium
CN113919923B (en) Live broadcast recommendation model training method, live broadcast recommendation method and related equipment
CN112269943B (en) Information recommendation system and method
CN110933504A (en) Video recommendation method, device, server and storage medium
CN113935251B (en) User behavior prediction model generation method and device and user behavior prediction method and device
CN112269942B (en) Method, device and system for recommending object and electronic equipment
CN111597444B (en) Searching method, searching device, server and storage medium
CN114491093A (en) Multimedia resource recommendation and object representation network generation method and device
CN113326385B (en) Target multimedia resource acquisition method and device, electronic equipment and storage medium
US20230362444A1 (en) Computing device and operating method thereof

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