CN114625893A - Media resource recall method, device, server and storage medium - Google Patents
Media resource recall method, device, server and storage medium Download PDFInfo
- Publication number
- CN114625893A CN114625893A CN202011457928.4A CN202011457928A CN114625893A CN 114625893 A CN114625893 A CN 114625893A CN 202011457928 A CN202011457928 A CN 202011457928A CN 114625893 A CN114625893 A CN 114625893A
- Authority
- CN
- China
- Prior art keywords
- media resource
- media
- vector
- resources
- graph
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/40—Information retrieval; Database structures therefor; File system structures therefor of multimedia data, e.g. slideshows comprising image and additional audio data
- G06F16/43—Querying
- G06F16/435—Filtering based on additional data, e.g. user or group profiles
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/40—Information retrieval; Database structures therefor; File system structures therefor of multimedia data, e.g. slideshows comprising image and additional audio data
- G06F16/48—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/40—Information retrieval; Database structures therefor; File system structures therefor of multimedia data, e.g. slideshows comprising image and additional audio data
- G06F16/48—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
- G06F16/483—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/40—Information retrieval; Database structures therefor; File system structures therefor of multimedia data, e.g. slideshows comprising image and additional audio data
- G06F16/48—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
- G06F16/489—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using time information
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Library & Information Science (AREA)
- Multimedia (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The disclosure relates to a media resource recall method, a media resource recall device, a server and a storage medium. The method comprises the following steps: acquiring at least one seed media resource corresponding to user information associated with a client; determining at least one candidate media resource corresponding to each seed media resource, taking a media resource vector of each media resource in a media resource set formed by the seed media resources and the candidate media resources as a vertex, and connecting every two vertexes to generate an undirected graph; and segmenting the undirected graph according to the attribute information of each edge in the undirected graph to obtain at least two sub-graphs, and determining the media resources to be recalled from the media resource set according to the distance between each media resource vector in each sub-graph and the central point vector of the corresponding sub-graph. According to the scheme, the problem that the recall frequency of the hot media resources is high but the recall frequency of the cold media resources is low in the related technology is solved, and the recall frequency of the cold media resources is improved while the diversity of the recalled media resources is improved.
Description
Technical Field
The present disclosure relates to computer technologies, and in particular, to a method, an apparatus, a server, and a storage medium for recalling a media resource.
Background
The development of computer and network technologies makes information more developed and more convenient to spread. Accordingly, how to select more targeted content from massive information and content to show to users becomes a problem of concern.
A media asset is a relatively special piece of information and content. The current internet media resource system generally comprises two stages of recall and sequencing; the recalling is responsible for acquiring a large number of media resources matched with the user information in the full library of media resources of the media resource system; in the related art, media resources are recalled mainly by a model-based method or a collaborative filtering-based method. However, the media assets recalled by both methods are poor in diversity, and are biased to recall the hot media assets, so that it is difficult to recall the cold media assets.
Disclosure of Invention
The present disclosure provides a media resource recall method, apparatus, server and storage medium, to at least solve the problem in the related art that the recall frequency of a hot media resource is high, but the recall frequency of a cold media resource is low.
The technical scheme of the embodiment of the disclosure is as follows:
according to a first aspect of an embodiment of the present disclosure, there is provided a media resource recall method, including:
responding to a media resource playing instruction triggered by a client, and acquiring at least one seed media resource corresponding to user information associated with the client, wherein the seed media resource is determined based on historical media resources associated with the user information and participating in value screening;
determining at least one candidate media resource corresponding to each seed media resource, taking a media resource vector of each media resource in a media resource set formed by the seed media resources and the candidate media resources as a vertex, connecting every two vertices to generate an undirected graph, wherein attribute information of an edge between every two vertices in the undirected graph is determined based on similarity between the corresponding media resource vectors;
segmenting the undirected graph according to attribute information of each edge in the undirected graph to obtain at least two sub-graphs, and determining recalled media resources from the media resource set according to the distance between each media resource vector in each sub-graph and a central point vector of the corresponding sub-graph; wherein each of the subgraphs contains a vector of media resources for at least one of the set of media resources.
Optionally, the step of obtaining at least one seed media resource corresponding to the user information associated with the client in response to the media resource playing instruction triggered by the client includes:
responding to a media resource playing instruction triggered by a client, and acquiring user information associated with the client;
and selecting at least one seed media resource corresponding to the user information from the recorded historical media resources entering a target media resource queue of the media resource system, wherein the historical media resources with media resource value parameters exceeding a first set threshold value are stored in the target media resource queue.
Optionally, the step of determining at least one candidate media resource corresponding to each seed media resource includes:
and retrieving the media resource vector of at least one candidate media resource similar to each seed media resource from a media resource vector library.
Optionally, the step of connecting every two vertexes to generate an undirected graph, where the media resource vector of each media resource in the media resource set formed by the seed media resource and the candidate media resource is used as a vertex, includes:
generating a vertex set based on a media resource vector of each media resource in a media resource set consisting of the seed media resource and the candidate media resource;
and respectively connecting the target vertex in the vertex set with all the vertices except the target vertex to generate the undirected graph.
Optionally, the step of segmenting the undirected graph according to the attribute information of each edge in the undirected graph to obtain at least two sub-graphs includes:
deleting the edge of the undirected graph, wherein the attribute information of the undirected graph is smaller than a second set threshold;
and traversing each vertex of the undirected graph, and determining at least two connected subgraphs in the undirected graph, wherein the at least two connected subgraphs are at least two subgraphs obtained by segmenting the undirected graph.
Optionally, the step of determining a recall media resource from the media resource set according to a distance between each media resource vector in each sub-graph and a center point vector of the corresponding sub-graph includes:
calculating a weighted average value of each media resource vector included in each sub-graph, and taking the weighted average value as a central point vector of each sub-graph;
calculating the distance between each media resource vector in each sub-graph and the central point vector of the corresponding sub-graph;
and when the distance is smaller than a third set threshold value, determining that the seed media resource or the candidate media resource corresponding to the distance smaller than the third set threshold value in the media resource set is the media resource to be recalled.
Optionally, the media resource vector is determined by a pre-trained delivery model;
the issuing model is obtained by training based on a positive sample formed by the media resources which participate in the value screening and are issued in the target media resource queue and a negative sample formed by the media resources which participate in the value screening and are not issued in the target media resource queue;
wherein, the media resources in the target media resource queue are determined by the following steps:
and determining a target recall media resource of which the media resource value parameter exceeds a set threshold value in the recall media resources corresponding to the user information, and adding the target recall media resource to a target media resource queue.
According to a second aspect of the embodiments of the present disclosure, there is provided a media resource recall apparatus including:
the system comprises an acquisition module, a selection module and a display module, wherein the acquisition module is configured to respond to a media resource playing instruction triggered by a client to acquire at least one seed media resource corresponding to user information associated with the client, and the seed media resource is determined based on historical media resources related to the user information and participating in value screening;
the determining module is configured to determine at least one candidate media resource corresponding to each seed media resource, take a media resource vector of each media resource in a media resource set formed by the seed media resources and the candidate media resources as a vertex, connect every two vertices to generate an undirected graph, and determine attribute information of an edge between every two vertices in the undirected graph based on similarity between the corresponding media resource vectors;
the partitioning module is configured to partition the undirected graph according to attribute information of each edge in the undirected graph to obtain at least two sub-graphs, and determine a recall media resource from the media resource set according to the distance between each media resource vector in each sub-graph and a central point vector of the corresponding sub-graph; wherein each of the subgraphs contains a vector of media resources for at least one of the set of media resources.
Optionally, the obtaining module is specifically configured to:
responding to a media resource playing instruction triggered by a client, and acquiring user information associated with the client;
and selecting at least one seed media resource corresponding to the user information from the recorded historical media resources entering a target media resource queue of the media resource system, wherein the historical media resources with media resource value parameters exceeding a first set threshold value are stored in the target media resource queue.
Optionally, the determining module includes: a retrieval submodule;
the retrieval submodule is configured to retrieve, from a media resource vector library, a media resource vector of at least one candidate media resource similar to each seed media resource.
Optionally, the determining module further includes: generating a submodule;
the generating submodule is configured to generate a vertex set based on a media resource vector of each media resource in a media resource set formed by the seed media resource and the candidate media resource;
and respectively connecting the target vertex in the vertex set with all the vertices except the target vertex to generate the undirected graph.
Optionally, the segmentation module includes: partitioning the sub-modules;
the segmentation submodule is configured to delete the side, in the undirected graph, of which the attribute information is smaller than a second set threshold;
and traversing each vertex of the undirected graph, and determining at least two connected subgraphs in the undirected graph, wherein the at least two connected subgraphs are at least two subgraphs obtained by segmenting the undirected graph.
Optionally, the segmentation module further includes: determining a submodule;
the determining sub-module is configured to calculate a weighted average value of media resource vectors included in each sub-graph, and take the weighted average value as a central point vector of each sub-graph;
calculating the distance between each media resource vector in each sub-graph and the central point vector of the corresponding sub-graph;
and when the distance is smaller than a third set threshold value, determining that the seed media resource or the candidate media resource corresponding to the distance smaller than the third set threshold value in the media resource set is the media resource to be recalled.
Optionally, the media resource vector is determined by a pre-trained delivery model;
the issuing model is obtained by training based on a positive sample formed by the media resources which participate in the value screening and are issued in the target media resource queue and a negative sample formed by the media resources which participate in the value screening and are not issued in the target media resource queue;
determining, by a second determining module, a media resource in the target media resource queue:
the second determining module is configured to determine a target recall media resource of which a media resource value parameter exceeds a set threshold in recall media resources corresponding to the user information, and add the target recall media resource to a target media resource queue.
According to a third aspect of the embodiments of the present disclosure, there is provided a server, including:
a processor;
a memory for storing executable instructions of the processor;
wherein the processor is configured to execute the instructions to implement the media resource recall method according to any one of the embodiments of the present disclosure.
According to a fourth aspect of the embodiments of the present disclosure, there is provided a storage medium, wherein instructions of the storage medium, when executed by a processor of a server, enable the server to perform a media asset recall method according to any one of the embodiments of the present disclosure.
According to a fifth aspect of embodiments of the present disclosure, there is provided a computer program product, wherein when the instructions in the computer program product are executed by a processor of a server, the method for recalling a media resource according to any embodiment of the present disclosure is implemented.
The technical scheme provided by the embodiment of the disclosure at least brings the following beneficial effects: by acquiring at least one seed media resource corresponding to user information associated with a client, it can be ensured that a plurality of seed media resources corresponding to the user information are acquired; by determining at least one candidate media resource corresponding to each seed media resource, even if the candidate media resource is a cold media resource, the candidate media resource can be acquired; taking a media resource vector of each media resource in a media resource set consisting of the seed media resource and the candidate media resource as a vertex, and connecting every two vertexes to generate an undirected graph; the method comprises the steps that an undirected graph is segmented according to attribute information of each side in the undirected graph to obtain at least two sub-graphs, and recalled media resources are determined from a media resource set according to the distance between each media resource vector in each sub-graph and a central point vector of the corresponding sub-graph, so that the diversity of the recalled media resources can be guaranteed, and the media resources can comprise hot media resources and cold media resources; therefore, the solution of the embodiment of the present disclosure can solve the problem that the recall frequency of the popular media resource is high but the recall frequency of the cold media resource is low in the related art, and improve the variety of the recalled media resources and the recall frequency of the cold media resource.
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 and are not to be construed as limiting the disclosure.
FIG. 1 is a flow chart illustrating a media asset recall method according to an exemplary embodiment.
FIG. 2 is a flow chart illustrating a media asset recall method according to an exemplary embodiment.
FIG. 3 is a flow chart illustrating a media asset recall method according to an exemplary embodiment.
FIG. 4 is an undirected graph containing 4 vertices, shown in accordance with an exemplary embodiment.
FIG. 5 is a flowchart illustrating a media asset recall method according to an exemplary embodiment.
FIG. 6 is a flow chart illustrating a media asset recall method according to an exemplary embodiment.
FIG. 7 is a flowchart illustrating the operation of a media asset system according to an exemplary embodiment.
FIG. 8 is a block diagram illustrating a media asset recall device, according to an exemplary embodiment.
FIG. 9 is a block diagram illustrating a server in accordance with an example embodiment.
Detailed Description
In order to make the technical solutions of the present disclosure better understood by those of ordinary skill in the art, the technical solutions in the embodiments of the present disclosure will be clearly and completely described below with reference to the accompanying drawings.
It should be noted that the terms "first," "second," and the like in the description and claims of the present disclosure and in the above-described drawings are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used is interchangeable under appropriate circumstances such that the embodiments of the disclosure described herein are capable of operation in sequences other than those illustrated or otherwise described herein. 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.
Fig. 1 is a flowchart illustrating a media resource recall method according to an exemplary embodiment, where the media resource recall method, as shown in fig. 1, may be performed by a media resource recall apparatus, which may be implemented by software and/or hardware and used in a server, and the media resource recall method includes the following steps.
In step S11, in response to the media resource playing instruction triggered by the client, at least one seed media resource corresponding to the user information associated with the client is obtained.
It should be noted that the media resources related in this embodiment may be plane resources, internet resources, or advertisement resources, and the like, which is not limited in this embodiment.
The client may also be referred to as a user side, which refers to a program for providing local services for a user; for example, the video playing program, the chat program, or the weather query program, etc., which are not limited in this embodiment. In an optional implementation manner of this embodiment, when the user opens the client or browses information in the client, the user may trigger a play instruction of the media resource.
The user information associated with the client may include an Identity (ID) of the user logging in the client, a user name or a mobile phone number of the user, and the like, which is not limited in this embodiment,
In an optional implementation manner of this embodiment, after receiving a media resource playing instruction triggered by a client, a server may further obtain at least one seed media resource corresponding to user information associated with the client; wherein the seed media asset is determined based on historical media assets associated with the user information that have participated in value screening (e.g., bidding). For example, at least one seed media asset may be selected from the historical media assets associated with the user's ID that last participated in bidding; at least one seed media resource can also be selected from the historical media resources which are associated with the mobile phone number of the user and participate in bidding last time, which is not limited in the embodiment.
For example, after receiving a media resource playing instruction triggered by the client, the server may obtain, according to a user name of a user logging in the client, 100 historical media resources related to the user name and participating in bidding last time in the media resource system, and determine the 100 historical media resources as a seed media resource corresponding to the user information related to the client.
In step S12, at least one candidate media resource corresponding to each seed media resource is determined, a media resource vector of each media resource in a media resource set formed by the seed media resources and the candidate media resources is used as a vertex, and every two vertices are connected to generate an undirected graph.
The media resource vector is an operation result obtained by operating the seed media resource or the candidate media resource by adopting a pre-trained issuing model. For example, the seed media resource or the candidate media resource is input into a pre-trained issuing model for prediction, so as to obtain a media resource vector corresponding to the seed media resource or the candidate media resource. In an optional implementation manner of this embodiment, the delivery model may be obtained by training based on a positive sample formed by the media resources that participate in the value screening and are delivered in the target media resource queue, and a negative sample formed by the media resources that participate in the value screening and are not delivered in the target media resource queue; wherein, the media resources in the target media resource queue can be determined by the following steps: and determining a target recall media resource of which the media resource value parameter exceeds a set threshold value in the recall media resource corresponding to the user information, and adding the target recall media resource to the target media resource queue.
Wherein the candidate media resources are similar to their corresponding seed media resources; alternatively, two media assets are determined to be similar if their media asset vectors are similar. For example, two media assets are determined to be similar when the distance of their media asset vectors is less than a set similarity threshold. For example, the seed media resource is a media resource of a living product a produced by a manufacturer M, and then the candidate media resource corresponding to the seed media resource may be a media resource of a living product B produced by the manufacturer M; the media resources may also be media resources of a living product a produced by the manufacturer N, which is not limited in this embodiment.
It should be further noted that the candidate media resources referred to in this embodiment may include both hot media resources and cold media resources of the seed media resources corresponding to the candidate media resources, and it is understood that, as long as media resources similar to the seed media resources can be obtained, parameters such as categories or values of the candidate media resources are not limited in this embodiment.
In an alternative implementation manner of this embodiment, media resource vectors of a plurality of candidate media resources similar to the media resource vector of each seed media resource may be retrieved in the media resource vector library. The media resource vector library comprises the association relation between the media resources and the corresponding media resource vectors. And searching candidate media resources in the media resource library based on the incidence relation between the media resource vector and the media resources in the media resource vector library to obtain a media resource set comprising the seed media resources and the candidate media resources. For example, the media resource vectors of 100 candidate media resources similar to the seed media resource a may be obtained in the media resource library according to the media resource vector of the seed media resource a; the media resource vectors of 200 candidate media resources similar to the seed media resource B may also be obtained in the media resource library, which is not limited in this embodiment. The media resource vector of the seed media resource is an operation result obtained by adopting a pre-trained issuing model to operate the seed media resource.
In an optional implementation manner of this embodiment, after determining at least one candidate media resource corresponding to each seed media resource, the seed media resources and the candidate media resources may be added to the media resource set, and further, each media resource vector in the media resource set may be used as a vertex, and further, no two vertices are connected to generate an undirected graph.
Wherein the attribute information of the edge between every two vertices in the undirected graph is determined based on similarity between vectors for the media resources; for example, if vertex a corresponds to media resource vector a and vertex B corresponds to media resource vector B, the attribute information of the edge between vertex a and vertex B is the similarity between media resource vector a and media resource vector B
Illustratively, if 10 seed media resources corresponding to the user information associated with the client are acquired, and 20 candidate media resources similar to each seed media resource are acquired in the media resource library, the 10 seed media resources and 200 candidate media resources may be added to the media resource set; furthermore, all media resource vectors in the media resource set can be used as each vertex of the undirected graph, every two vertices are connected to generate the undirected graph, and the attribute information of the edge between every two vertices is the similarity between the corresponding media resource vectors.
In step S13, the undirected graph is segmented according to the attribute information of each edge in the undirected graph to obtain at least two sub-graphs, and a media resource to be recalled is determined from the media resource set according to the distance between each media resource vector in each sub-graph and the center point vector of the corresponding sub-graph.
Wherein each subgraph contains a media resource vector for at least one media resource in the set of media resources; for example, the media resource vector of at least one seed media resource in the media resource set, the media resource vector of the candidate media resource, or both the media resource vector of at least one seed media resource in the media resource set and the media resource vector of at least one candidate media resource may be included, which is not limited in this embodiment.
In an optional implementation manner of this embodiment, after a media resource vector of each media resource in a media resource set is used as a vertex, and no two vertices are connected to generate an undirected graph, the generated undirected graph may be divided according to attribute information of each edge in the undirected graph, for example, edges with the same similarity (attribute information of the edge) or the difference smaller than a set threshold (e.g., 0.01) in the undirected graph may be deleted and divided into a sub-graph.
Further, a central point vector of each sub-graph may be calculated, for example, in the sub-graph a obtained by segmentation, all media resource vectors included in the sub-graph a may be added, and then an average value is obtained, and the average value may be used as the central point vector of the sub-graph a; after the central point vector of each sub-graph is obtained through calculation, media resources to be recalled can be determined from the media resource set according to the distance between the media resource vector in each sub-graph and the central point vector of the corresponding sub-graph; for example, in the above example, after the center point vector of the sub-graph a is obtained through calculation, a distance between each media resource vector in the sub-graph a and the center point vector may be calculated, and when the distance value is smaller than a third set threshold, the seed media resource or the candidate media resource in the media resource set corresponding to the distance smaller than the third set threshold is determined as the media resource to be recalled.
The third set threshold may be a numerical value such as 0.1, 0.2, or 0.05, which is not limited in this embodiment.
In the related art, media resources are recalled mainly by a model-based method or a collaborative filtering-based method. Wherein the model-based method: and taking the media resource issued to the client as a negative sample, taking the media resource clicked by the user at the client as a positive sample, training the model, and determining the media resource to be recalled according to the model obtained by training. The method based on collaborative filtering comprises the following steps: and mining according to association rules between users and between media resources, so as to obtain the recalled media resources. The method in the related art is biased to recall the hot media resources with higher media resource value parameters, and is difficult to recall the cold media resources with lower media resource value parameters, so that the recalled media resources have poor diversity.
According to the scheme of the embodiment, by acquiring at least one seed media resource corresponding to the user information associated with the client, a plurality of seed media resources corresponding to the user information can be ensured to be acquired; by determining at least one candidate media resource corresponding to each seed media resource, even if the candidate media resource is a cold media resource (a media resource with a low media resource value parameter), the candidate media resource can be acquired; taking a media resource vector of each media resource in a media resource set consisting of the seed media resource and the candidate media resource as a vertex, and connecting every two vertexes to generate an undirected graph; the method comprises the steps that an undirected graph is divided according to attribute information of each side in the undirected graph to obtain at least two sub-graphs, and recalled media resources are determined from a media resource set according to the distance between each media resource vector in each sub-graph and a central point vector of the corresponding sub-graph, so that the diversity of the recalled media resources (including hot media resources and cold media resources) can be ensured; therefore, the solution of the embodiment of the present disclosure can solve the problem that the recall frequency of the popular media resource is high but the recall frequency of the cold media resource is low in the related art, and improve the variety of the recalled media resources and the recall frequency of the cold media resource.
Fig. 2 is a flowchart illustrating a media resource recall method according to an exemplary embodiment, which is a further refinement of the above technical solution, and the technical solution in the embodiment may be combined with various alternatives in one or more of the above embodiments. As shown in fig. 2, the media asset recall method includes the following steps.
In step S21, in response to the media resource playing instruction triggered by the client, user information associated with the client is obtained; and selecting at least one seed media resource corresponding to the user information from the recorded historical media resources of the target media resource queue entering the media resource system.
The target media resource queue stores historical media resources with media resource value parameters exceeding a first set threshold. Wherein, the media resource value parameter is CPM (Cost Per thousand impressions), that is, each media resource represents 1000 times of Cost; the first set threshold may be an amount of money such as 10 yuan, 100 yuan, or 200 yuan, which is not limited in this embodiment.
In an optional implementation manner of this embodiment, after receiving a media resource playing instruction triggered by a client, a server may further obtain user information associated with the client; for example, a user ID, a user name, an account number, or the like of the login client may be acquired, which is not limited in this embodiment.
Further, at least one seed media resource corresponding to the user information may be selected from historical media resources recorded in a target media resource queue of the media resource system. For example, if the time when the server receives the media resource play instruction triggered by the client is 7/month and 2/2020, the historical media resources recorded in the target media resource queue may be 100 media resources associated with the user information and participating in bidding on 7/month and 1/2020. It should be noted that the number of the selected historical media resources may be an empirical value related to the application scenario, and the more the selected historical media resources, the larger the selection space of the seed media resource.
In step S22, a media resource vector of at least one candidate media resource similar to each seed media resource is retrieved from the media resource vector library.
In an optional implementation manner of this embodiment, after at least one seed media resource corresponding to the user information is selected from the recorded historical media resources of the target media resource queue entering the media resource system, a media resource vector of at least one candidate media resource similar to each seed media resource may be retrieved from a media resource vector library.
The media resource vector library stores media resource vectors of all media resources; all media resources are media resources for all products, individuals or companies, and are not limited in this embodiment. It should be noted that the media resource vectors of all the media resources can be obtained by respectively calculating each media resource by adopting a pre-trained issuing model; for example, each media resource may be sequentially input into a pre-trained delivery model for prediction, so as to obtain a media resource vector of each media resource, and the media resource vectors are stored in a media resource vector library.
In an optional implementation manner of this embodiment, a media resource vector of at least one candidate media resource similar to each seed media resource may be retrieved from a media resource vector library by using a vector retrieval algorithm such as HNSW (Hierarchical Navigable Small World) or LSH (local Sensitive Hashing).
In step S23, a media resource vector of each media resource in the media resource set composed of the seed media resource and the candidate media resource is used as a vertex, and every two vertices are connected to generate an undirected graph.
In step S24, the undirected graph is segmented according to the attribute information of each edge in the undirected graph to obtain at least two sub-graphs, and a media resource to be recalled is determined from the media resource set according to the distance between each media resource vector in each sub-graph and the center point vector of the corresponding sub-graph.
According to the scheme of the embodiment, after at least one seed media resource corresponding to the user information is selected from the recorded historical media resources of the target media resource queue entering the media resource system, a media resource vector of at least one candidate media resource similar to each seed media resource can be further retrieved from a media resource vector library, and the retrieved media resource vector includes both hot media resources and cold media resources, so that the candidate media resources with diversity can be ensured to be obtained, even the cold media resources can be obtained, and a basis is provided for subsequently improving the diversity of media resources to be recalled.
Fig. 3 is a flowchart illustrating a media resource recall method according to an exemplary embodiment, which is a further refinement of the above technical solution, and the technical solution in the embodiment may be combined with various alternatives in one or more of the above embodiments. As shown in fig. 3, the media asset recall method includes the following steps.
In step S31, in response to the media resource playing instruction triggered by the client, user information associated with the client is obtained; and selecting at least one seed media resource corresponding to the user information from the recorded historical media resources of the target media resource queue entering the media resource system.
In step S32, a media resource vector of at least one candidate media resource similar to each seed media resource is retrieved from the media resource vector library.
In step S33, a media resource vector of each media resource in the media resource set composed of the seed media resource and the candidate media resource is used as a vertex, and every two vertices are connected to generate an undirected graph. In an optional implementation manner of this embodiment, taking a media resource vector of each media resource in a media resource set composed of a seed media resource and a candidate media resource as a vertex, and connecting every two vertices to generate an undirected graph may include: generating a vertex set based on a media resource vector of each media resource in a media resource set consisting of the seed media resource and the candidate media resource; and respectively connecting the target vertex in the vertex set with all the vertexes except the target vertex to generate an undirected graph. The target vertex may be any media resource vector in the media resource set, which is not limited in this embodiment.
In an optional implementation manner, after the candidate media resources corresponding to all the seed media resources are obtained, a media resource vector of each media resource in the media resource set may be further used as a vertex set, that is, the media resource vector of each media resource in the media resource set is used as each vertex of an undirected graph, where the media resource set includes all the seed media resources and the candidate media resources corresponding to all the seed media resources; and taking the similarity of the media resource vector between every two media resources in the media resource set as the attribute information of each edge, namely taking the similarity of the media resource vector between every two media resources in the media resource set as the attribute information of each edge of the undirected graph.
For example, if the media resource set includes 4 media resources (the seed media resource and the candidate media resource) in total, the generated undirected graph can be as shown in fig. 4, which includes 4 vertices (1, 2, 3, and 4) and 6 edges (5, 6, 7, 8, 9, and 10), wherein the vertices can represent media resource vectors of the 4 media resources; the attribute information of the edge may represent similarity of media asset vectors between every two media assets of the 4 media assets.
In step S34, a side of the undirected graph whose attribute information is smaller than a second set threshold is deleted; and traversing each vertex of the undirected graph, and determining at least two connected subgraphs in the undirected graph, wherein the at least two connected subgraphs are at least two subgraphs obtained by segmenting the undirected graph.
In an optional implementation manner of this embodiment, after a media resource vector of each media resource in a media resource set composed of a seed media resource and a candidate media resource is used as a vertex and every two vertices are connected to generate an undirected graph, an edge whose attribute information (similarity) is smaller than a second set threshold in the undirected graph may be further deleted; the second set threshold may be 0.2, 0.3, or 0.5, which is not limited in this embodiment.
For example, if the similarity between the first media resource vector and the second media resource vector is 0.1 (the second set threshold is 0.2), the edge corresponding to the media resource vector similarity (attribute information) between the first media resource vector and the second media resource vector in the undirected graph may be deleted, and at this time, the first media resource corresponding to the first media resource vector and the second media resource corresponding to the second media resource vector are not connected. The advantage of this arrangement is that it can provide basis for subsequent segmentation undirected graphs, and reduce the amount of computation.
It can be understood that after the edges with the similarity smaller than the second set threshold in the undirected graph are deleted, the undirected graph can be converted into a plurality of connected undirected graphs from one connected graph, and at this time, each vertex of the undirected graph can be traversed, so that a plurality of connected subgraphs included in the undirected graph are determined. It should be noted that if a plurality of connected undirected graphs (for example, 2 undirected graphs) are not obtained after deleting the side of the undirected graph whose similarity is smaller than the second set threshold, the second set threshold may be adjusted at this time, and for example, the second set threshold may be adjusted to be smaller, for example, the second set threshold is adjusted from 0.2 to 0.15, until at least two connected undirected graphs are obtained.
Illustratively, processing the undirected graph as shown in fig. 4 removes edges in the undirected graph having a similarity less than a second set threshold, and further traversal of each vertex of the undirected graph may determine two subgraphs, a first subgraph comprising vertex 1 and a second subgraph comprising vertex 2, vertex 3 and vertex 4, assuming edges 5, 6 and 7 are removed.
In step S35, a recall media resource is determined from the set of media resources based on the distance of each media resource vector in each sub-graph from the center point vector of the corresponding sub-graph.
In the scheme of the embodiment, a vertex set is generated based on a media resource vector of each media resource in a media resource set consisting of a seed media resource and a candidate media resource; respectively connecting the target vertex in the vertex set with all the vertexes except the target vertex to generate an undirected graph; deleting edges of which the attribute information is smaller than a second set threshold in the undirected graph; traversing each vertex of the undirected graph, determining at least two connected subgraphs in the undirected graph, wherein the at least two connected subgraphs are the at least two subgraphs obtained by segmenting the undirected graph, the undirected graph can be rapidly segmented, the efficiency of media resource recall is improved, a media resource playing instruction triggered by a client can be rapidly responded, and the time for issuing the media resource by a media resource system is reduced.
Fig. 5 is a flowchart illustrating a media resource recall method according to an exemplary embodiment, which is a further refinement of the above technical solution, and the technical solution in the embodiment may be combined with various alternatives in one or more of the above embodiments. As shown in fig. 5, the media asset recall method includes the following steps.
In step S51, in response to the media resource playing instruction triggered by the client, at least one seed media resource corresponding to the user information associated with the client is obtained.
In step S52, at least one candidate media resource corresponding to each seed media resource is determined, a media resource vector of each media resource in a media resource set formed by the seed media resources and the candidate media resources is used as a vertex, and every two vertices are connected to generate an undirected graph.
In step S53, a side of the undirected graph whose attribute information is smaller than a second set threshold is deleted; and traversing each vertex of the undirected graph, and determining at least two connected subgraphs in the undirected graph, wherein the at least two connected subgraphs are at least two subgraphs obtained by segmenting the undirected graph.
In step S54, calculating a weighted average of media resource vectors included in each sub-graph, and taking the weighted average as a central point vector of each sub-graph; calculating the distance between each media resource vector in each sub-graph and the central point vector of the corresponding sub-graph; and when the distance is smaller than a third set threshold value, determining the seed media resource or the candidate media resource corresponding to the distance smaller than the third set threshold value in the media resource set as the media resource to be recalled.
In an optional implementation manner of this embodiment, after the generated undirected graph is segmented to obtain at least two sub-graphs, a weighted average of media resource vectors included in each sub-graph may be further calculated, and the calculated weighted average is used as a center point vector of each sub-graph.
For example, if the sub-graph a obtained by partitioning includes 10 media resource vectors (including a seed media resource vector and a candidate media resource vector), different weights may be set for the 10 media resource vectors, for example, the weight of the media resource vector at the center of the sub-graph is set to 1, the weight of the media resource vector at the boundary of the sub-graph is set to 0.8, and the like, which is not limited in this embodiment; further, a weighted average of the 10 media resource vectors can be obtained through calculation, and the weighted average of the 10 media resource vectors is determined as a central point vector of the sub-graph a.
Further, distances between all media resource vectors in the sub-graph a and the calculated central point vector of the sub-graph a may be calculated, for example, distances between 10 media resource vectors in the sub-graph a and the central point vector are respectively calculated to obtain 10 distance values; further, whether the distance is smaller than a third set threshold value is determined; the third set threshold may be a numerical value such as 0.1, 0.2, or 0.05, which is not limited in this embodiment. For example, if all of the above 10 distance values are smaller than the third set threshold, the 10 media resource vectors in sub-graph a may be determined as the media resources to be recalled.
In another optional implementation manner of this embodiment, a median or an average of media resource vectors included in each sub-graph may also be calculated, and the calculated median or average is used as a center point vector of each sub-graph; calculating the distance between each media resource vector in each sub-graph and the central point vector of the corresponding sub-graph; and when the distance is smaller than a third set threshold value, determining the seed media resource or the candidate media resource corresponding to the distance smaller than the third set threshold value in the media resource set as the media resource to be recalled. In the scheme of this embodiment, a weighted average of media resource vectors included in each sub-graph is calculated, and the weighted average is used as a central point vector of each sub-graph; calculating the distance between each media resource vector in each sub-graph and the central point vector of the corresponding sub-graph; when the distance is smaller than the third set threshold, it is determined that the seed media resource or the candidate media resource in the media resource set corresponding to the distance smaller than the third set threshold is the media resource to be recalled, so that the diversity of the media resource to be recalled can be improved (the popular media resource can be obtained, and also the popular media resource can be obtained).
In order to make those skilled in the art better understand the media resource recall method related to the present embodiment, a specific example is used for description below, and it should be noted that in the present embodiment, an advertisement is used as a media resource for description; as shown in fig. 6, the specific process includes:
in step S61, a user seed advertisement is obtained.
Recording each advertisement requested to enter into the bidding by the advertisement system, and randomly keeping M advertisements entering into the bidding queue by each user, wherein M can be any value, and is not limited in the embodiment.
In step S62, similar advertisements are retrieved.
And acquiring seed advertisement vectors through the issuing model, and searching at least one advertisement similar to each seed advertisement by using vector search algorithms such as HNSW (human neural network) or LSH (least squares).
In step S63, an undirected graph is generated.
In an advertisement set consisting of various sub-advertisements and at least one advertisement similar to the various sub-advertisements, the similarity of every two advertisement vectors is calculated, an undirected graph is generated, and edges with the similarity smaller than a set threshold value are deleted.
In step S64, the undirected graph is minimally cut into at least two subgraphs.
And adopting a minimum cutting algorithm to cut the undirected graph into at least two sub-graphs, wherein each sub-graph is taken as a focus point, and the central point vector is the weighted average of all the dimensionality vectors of all the points of the sub-graph. It should be noted that the minimal cut algorithm in this embodiment may also be replaced by a clustering algorithm, which is not limited in this embodiment.
In step S65, the advertisements are filtered.
At least one advertisement vector is selected for each sub-graph that is closest to the center point.
In step S66, the recall result is returned.
According to the scheme of the embodiment, the problem of diversity can be solved to a certain extent by using the central point of the advertisement subgraph as the interest point; by adopting a minimum cut algorithm and a vector retrieval algorithm, the operation efficiency can be improved; the advertisement entering the bidding queue is selected as the seed point by the seed point, so that sufficient diversity is ensured, and the problems of slow feedback time and insufficient diversity caused by the fact that the user behavior is used as the seed point in the related technology are solved.
Fig. 7 is a flowchart illustrating an operation of an advertisement system according to an exemplary embodiment, and referring to fig. 7, the operation of the advertisement system mainly includes the following steps.
In step S71, recall.
According to the advertisement recalling method related to the embodiments, the advertisement satisfying the user request is recalled.
In step S72, sorting is performed.
And performing CPM estimation on each recalled advertisement according to the user information, the advertisement bid and the advertisement information.
In step S73, the bid is placed.
And filtering low-value advertisements according to the CPM estimation result, and screening out at least one high-value advertisement.
In step S74, the program is issued.
And sending the at least one high-value advertisement obtained by screening to the client, and displaying the delivered advertisement at the client.
In this embodiment, in the bidding and issuing stages, a positive sample formed by the advertisement which participates in bidding and is issued in the target advertisement queue and a negative sample formed by the advertisement which participates in bidding and is not issued in the target advertisement queue are respectively used to train the issuing model. The issuing model related in the embodiment is supervised training, and the value of the advertisement can be considered; meanwhile, the issuing model is a server model, so that the problem of slow feedback of the client model can be avoided, and the advertisement playing request triggered by the client can be fed back quickly.
FIG. 8 is a block diagram illustrating a media asset recall apparatus according to an exemplary embodiment. Referring to fig. 8, the apparatus includes an acquisition module 81, a determination module 82 and a segmentation module 83.
An obtaining module 81 configured to, in response to a media resource playing instruction triggered by a client, obtain at least one seed media resource corresponding to user information associated with the client, where the seed media resource is determined based on a historical media resource associated with the user information and participating in value screening;
a determining module 82 configured to determine at least one candidate media resource corresponding to each seed media resource, take a media resource vector of each media resource in a media resource set formed by the seed media resources and the candidate media resources as a vertex, connect every two vertices to generate an undirected graph, and determine attribute information of an edge between every two vertices in the undirected graph based on similarity between the corresponding media resource vectors;
a partitioning module 83 configured to partition the undirected graph according to attribute information of each edge in the undirected graph to obtain at least two sub-graphs, and determine a media resource to be recalled from the media resource set according to a distance between each media resource vector in each sub-graph and a center point vector of the corresponding sub-graph; wherein each subgraph contains at least one media resource of the set of media resources.
According to the scheme of the embodiment, the acquisition module responds to a media resource playing instruction triggered by the client to acquire at least one seed media resource corresponding to the user information associated with the client; determining at least one candidate media resource corresponding to each seed media resource through a determining module, taking a media resource vector of each media resource in a media resource set formed by the seed media resources and the candidate media resources as a vertex, and connecting every two vertices to generate an undirected graph; the division module divides the undirected graph according to the attribute information of each side in the undirected graph to obtain at least two sub-graphs, and determines the recall media resources from the media resource set according to the distance between each media resource vector in each sub-graph and the central point vector of the corresponding sub-graph, so that the problems of high recall frequency of hot media resources and low recall frequency of cold media resources in the related art are solved, and the recall frequency of the cold media resources is improved while the diversity of the recalled media resources is improved.
Optionally, the obtaining module 81 is specifically configured to:
responding to a media resource playing instruction triggered by a client, and acquiring user information associated with the client;
selecting at least one seed media resource corresponding to the user information from the recorded historical media resources entering a target media resource queue of the media resource system, wherein the historical media resources with media resource value parameters exceeding a first set threshold value are stored in the target media resource queue.
Optionally, the determining module 82 includes: a retrieval submodule;
and the retrieval submodule is configured to retrieve the media resource vector of at least one candidate media resource similar to each seed media resource from the media resource vector library.
Optionally, the determining module 82 further includes: generating a submodule;
the generating submodule is configured to generate a vertex set based on the media resource vector of each media resource in the media resource set formed by the seed media resource and the candidate media resource;
and respectively connecting the target vertex in the vertex set with all the vertices except the target vertex to generate an undirected graph.
Optionally, the segmentation module 83 includes: partitioning the sub-modules;
the segmentation submodule is configured to delete the side of the undirected graph, of which the attribute information is smaller than a second set threshold;
and traversing each vertex of the undirected graph, and determining at least two connected subgraphs in the undirected graph, wherein the at least two connected subgraphs are at least two subgraphs obtained by segmenting the undirected graph.
Optionally, the segmentation module 83 further includes: determining a submodule;
the determining sub-module is configured to calculate a weighted average value of media resource vectors included in each sub-graph, and the weighted average value is used as a central point vector of each sub-graph;
calculating the distance between each media resource vector in each sub-graph and the central point vector of the corresponding sub-graph;
and when the distance is smaller than a third set threshold value, determining the seed media resource or the candidate media resource corresponding to the distance smaller than the third set threshold value in the media resource set as the media resource to be recalled.
Optionally, the issuing model is obtained by training based on a positive sample formed by the media resources which participate in the bidding and are issued in the target media resource queue and a negative sample formed by the media resources which participate in the bidding and are not issued in the target media resource queue;
wherein, the media resource vector is determined by a pre-trained issuing model;
the issuing model is obtained by training based on a positive sample formed by the media resources which participate in the value screening and are issued in the target media resource queue and a negative sample formed by the media resources which participate in the value screening and are not issued in the target media resource queue;
determining the media resources in the target media resource queue through a second determination module:
and the second determining module is configured to determine a target recall media resource of which the media resource value parameter exceeds a set threshold value in the recall media resource corresponding to the user information, and add the target recall media resource to the target media resource queue.
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.
Fig. 9 is a block diagram illustrating a configuration of a server according to an example embodiment. As shown in fig. 9, the server includes a processor 91; a Memory 92 for storing executable instructions of the processor 91, the Memory 92 may include a RAM (Random Access Memory) and a ROM (Read-Only Memory); wherein the processor 91 is configured to execute instructions to implement the media asset recall method described above;
the method comprises the steps that in response to a media resource playing instruction triggered by a client, at least one seed media resource corresponding to user information associated with the client is obtained, wherein the seed media resource is determined based on historical media resources which are associated with the user information and participate in value screening;
determining at least one candidate media resource corresponding to each seed media resource, taking a media resource vector of each media resource in a media resource set formed by the seed media resources and the candidate media resources as a vertex, connecting every two vertices to generate an undirected graph, and determining attribute information of an edge between every two vertices in the undirected graph based on similarity between the corresponding media resource vectors;
dividing the undirected graph according to the attribute information of each side in the undirected graph to obtain at least two sub-graphs, and determining media resources to be recalled from the media resource set according to the distance between each media resource vector in each sub-graph and the central point vector of the corresponding sub-graph; wherein each subgraph contains at least one media resource of the set of media resources.
In an exemplary embodiment, there is also provided a storage medium comprising instructions, such as a memory 92 storing executable instructions, which are executable by a processor 91 of a server (server or smart terminal) to perform the media asset recall method described above.
Alternatively, the storage medium may be a non-transitory computer readable storage medium, for example, the non-transitory 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, a computer program product is also provided, in which instructions, when executed by a processor of a server, implement the above-described media resource recall method.
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 (10)
1. A method for recalling a media resource, comprising:
responding to a media resource playing instruction triggered by a client, and acquiring at least one seed media resource corresponding to user information associated with the client, wherein the seed media resource is determined based on historical media resources associated with the user information and participating in value screening;
determining at least one candidate media resource corresponding to each seed media resource, taking a media resource vector of each media resource in a media resource set formed by the seed media resources and the candidate media resources as a vertex, connecting every two vertices to generate an undirected graph, wherein attribute information of an edge between every two vertices in the undirected graph is determined based on similarity between the corresponding media resource vectors;
segmenting the undirected graph according to attribute information of each edge in the undirected graph to obtain at least two sub-graphs, and determining recalled media resources from the media resource set according to the distance between each media resource vector in each sub-graph and a central point vector of the corresponding sub-graph; wherein each of the subgraphs contains a vector of media resources for at least one of the set of media resources.
2. The method according to claim 1, wherein the step of obtaining at least one seed media resource corresponding to the user information associated with the client in response to a media resource playing instruction triggered by the client comprises:
responding to a media resource playing instruction triggered by a client, and acquiring user information associated with the client;
and selecting at least one seed media resource corresponding to the user information from the recorded historical media resources entering a target media resource queue of the media resource system, wherein the historical media resources with media resource value parameters exceeding a first set threshold value are stored in the target media resource queue.
3. The method of claim 1, wherein the step of determining at least one candidate media resource corresponding to each seed media resource comprises:
and retrieving the media resource vector of at least one candidate media resource similar to each seed media resource from a media resource vector library.
4. The method according to claim 3, wherein the step of generating an undirected graph by connecting every two vertexes by using a media resource vector of each media resource in the media resource set composed of the seed media resource and the candidate media resource as a vertex comprises:
generating a vertex set based on a media resource vector of each media resource in a media resource set consisting of the seed media resource and the candidate media resource;
and respectively connecting the target vertex in the vertex set with all the vertices except the target vertex to generate the undirected graph.
5. The method of claim 4, wherein the step of segmenting the undirected graph according to the attribute information of each edge in the undirected graph to obtain at least two subgraphs comprises:
deleting the edge of the undirected graph, wherein the attribute information of the undirected graph is smaller than a second set threshold;
and traversing each vertex of the undirected graph, and determining at least two connected subgraphs in the undirected graph, wherein the at least two connected subgraphs are at least two subgraphs obtained by segmenting the undirected graph.
6. The method of claim 1, wherein the step of determining a recall media resource from the set of media resources based on a distance of each media resource vector in each of the sub-graphs from a center point vector of the corresponding sub-graph comprises:
calculating a weighted average value of each media resource vector included in each sub-graph, and taking the weighted average value as a central point vector of each sub-graph;
calculating the distance between each media resource vector in each sub-graph and the central point vector of the corresponding sub-graph;
and when the distance is smaller than a third set threshold value, determining that the seed media resource or the candidate media resource corresponding to the distance smaller than the third set threshold value in the media resource set is the media resource to be recalled.
7. The method of any of claims 1-6, wherein the media asset vector is determined by a pre-trained delivery model;
the issuing model is obtained by training based on a positive sample formed by the media resources which participate in the value screening and are issued in the target media resource queue and a negative sample formed by the media resources which participate in the value screening and are not issued in the target media resource queue;
wherein, the media resources in the target media resource queue are determined by the following steps:
and determining a target recall media resource of which the media resource value parameter exceeds a set threshold value in the recall media resources corresponding to the user information, and adding the target recall media resource to a target media resource queue.
8. A media asset recall apparatus, comprising:
the system comprises an acquisition module, a selection module and a display module, wherein the acquisition module is configured to respond to a media resource playing instruction triggered by a client to acquire at least one seed media resource corresponding to user information associated with the client, and the seed media resource is determined based on historical media resources related to the user information and participating in value screening;
the determining module is configured to determine at least one candidate media resource corresponding to each seed media resource, take a media resource vector of each media resource in a media resource set formed by the seed media resources and the candidate media resources as a vertex, connect every two vertices to generate an undirected graph, and determine attribute information of an edge between every two vertices in the undirected graph based on similarity between the corresponding media resource vectors;
the partitioning module is configured to partition the undirected graph according to attribute information of each edge in the undirected graph to obtain at least two sub-graphs, and determine a recall media resource from the media resource set according to the distance between each media resource vector in each sub-graph and a central point vector of the corresponding sub-graph; wherein each of the subgraphs contains a vector of media resources for at least one of the set of media resources.
9. A server, comprising:
a processor;
a memory for storing executable instructions of the processor;
wherein the processor is configured to execute the instructions to implement the media asset recall method of any of claims 1-7.
10. A storage medium having instructions thereon that, when executed by a processor of a server, enable the server to perform a media asset recall method according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011457928.4A CN114625893A (en) | 2020-12-10 | 2020-12-10 | Media resource recall method, device, server and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011457928.4A CN114625893A (en) | 2020-12-10 | 2020-12-10 | Media resource recall method, device, server and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114625893A true CN114625893A (en) | 2022-06-14 |
Family
ID=81895612
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011457928.4A Pending CN114625893A (en) | 2020-12-10 | 2020-12-10 | Media resource recall method, device, server and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114625893A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115291503A (en) * | 2022-09-29 | 2022-11-04 | 北京达佳互联信息技术有限公司 | Recommendation method, recommendation device, electronic device and storage medium |
-
2020
- 2020-12-10 CN CN202011457928.4A patent/CN114625893A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115291503A (en) * | 2022-09-29 | 2022-11-04 | 北京达佳互联信息技术有限公司 | Recommendation method, recommendation device, electronic device and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11734725B2 (en) | Information sending method, apparatus and system, and computer-readable storage medium | |
| US20170065889A1 (en) | Identifying And Extracting Video Game Highlights Based On Audio Analysis | |
| CN111026971A (en) | Content pushing method and device and computer storage medium | |
| CN111681067A (en) | Long-tail commodity recommendation method and system based on graph attention network | |
| JP7480345B2 (en) | Area division method, device, electronic device, and computer program | |
| CN105718566B (en) | Intelligent music recommendation system | |
| KR20120024772A (en) | Method and apparatus for providing moving image advertisements | |
| CN113535991B (en) | Multimedia resource recommendation method and device, electronic equipment and storage medium | |
| Unger et al. | Inferring contextual preferences using deep auto-encoding | |
| KR101922725B1 (en) | Artificial intelligence project matching method, apparatus and program | |
| CN111626767B (en) | Resource data issuing method, device and equipment | |
| CN111327930A (en) | Method and device for acquiring target object, electronic equipment and storage medium | |
| KR20170036874A (en) | Method and apparatus for recommendation of social event based on users preference | |
| CN114625893A (en) | Media resource recall method, device, server and storage medium | |
| CN109451334A (en) | User, which draws a portrait, generates processing method, device and electronic equipment | |
| CN112036987B (en) | Method and device for determining recommended commodity | |
| CN109829593B (en) | Credit determining method and device for target object, storage medium and electronic device | |
| CN112115354A (en) | Information processing method, information processing apparatus, server, and storage medium | |
| US20220374474A1 (en) | Recommending content to subscribers | |
| Seychell et al. | Ranking regions of visual saliency in rgb-d content | |
| CN112749332B (en) | Data processing method, device and computer readable medium | |
| CN114329055A (en) | Search recommendation method and recommendation device, electronic device and storage medium | |
| CN110992109B (en) | Real estate customer analysis method, apparatus and storage medium based on association rule | |
| CN113486211A (en) | Account identification method and device, electronic equipment, storage medium and program product | |
| KR101944093B1 (en) | Project applicant evaluating method for artificial intelligence project matching |
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 |