Nothing Special   »   [go: up one dir, main page]

CN115033777B - Data recommendation method, electronic device and storage medium - Google Patents

Data recommendation method, electronic device and storage medium Download PDF

Info

Publication number
CN115033777B
CN115033777B CN202210436960.7A CN202210436960A CN115033777B CN 115033777 B CN115033777 B CN 115033777B CN 202210436960 A CN202210436960 A CN 202210436960A CN 115033777 B CN115033777 B CN 115033777B
Authority
CN
China
Prior art keywords
recommendation
sample
target
resource
recommended
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
CN202210436960.7A
Other languages
Chinese (zh)
Other versions
CN115033777A (en
Inventor
王国瑞
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.)
Tencent Music Entertainment Technology Shenzhen Co Ltd
Original Assignee
Tencent Music Entertainment Technology Shenzhen 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 Tencent Music Entertainment Technology Shenzhen Co Ltd filed Critical Tencent Music Entertainment Technology Shenzhen Co Ltd
Priority to CN202210436960.7A priority Critical patent/CN115033777B/en
Publication of CN115033777A publication Critical patent/CN115033777A/en
Application granted granted Critical
Publication of CN115033777B publication Critical patent/CN115033777B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/90—Details of database functions independent of the retrieved data types
    • G06F16/95—Retrieval from the web
    • G06F16/953—Querying, e.g. by the use of web search engines
    • G06F16/9535—Search customisation based on user profiles and personalisation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/90—Details of database functions independent of the retrieved data types
    • G06F16/95—Retrieval from the web
    • G06F16/953—Querying, e.g. by the use of web search engines
    • G06F16/9538—Presentation of query results
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
    • H04N21/20—Servers specifically adapted for the distribution of content, e.g. VOD servers; Operations thereof
    • H04N21/25—Management operations performed by the server for facilitating the content distribution or administrating data related to end-users or client devices, e.g. end-user or client device authentication, learning user preferences for recommending movies
    • H04N21/251—Learning process for intelligent management, e.g. learning user preferences for recommending movies
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
    • H04N21/20—Servers specifically adapted for the distribution of content, e.g. VOD servers; Operations thereof
    • H04N21/25—Management operations performed by the server for facilitating the content distribution or administrating data related to end-users or client devices, e.g. end-user or client device authentication, learning user preferences for recommending movies
    • H04N21/258—Client or end-user data management, e.g. managing client capabilities, user preferences or demographics, processing of multiple end-users preferences to derive collaborative data
    • H04N21/25866—Management of end-user data
    • H04N21/25891—Management of end-user data being end-user preferences
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
    • H04N21/40—Client devices specifically adapted for the reception of or interaction with content, e.g. set-top-box [STB]; Operations thereof
    • H04N21/45—Management operations performed by the client for facilitating the reception of or the interaction with the content or administrating data related to the end-user or to the client device itself, e.g. learning user preferences for recommending movies, resolving scheduling conflicts
    • H04N21/4508—Management of client data or end-user data
    • H04N21/4532—Management of client data or end-user data involving end-user characteristics, e.g. viewer profile, preferences
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
    • H04N21/40—Client devices specifically adapted for the reception of or interaction with content, e.g. set-top-box [STB]; Operations thereof
    • H04N21/45—Management operations performed by the client for facilitating the reception of or the interaction with the content or administrating data related to the end-user or to the client device itself, e.g. learning user preferences for recommending movies, resolving scheduling conflicts
    • H04N21/466—Learning process for intelligent management, e.g. learning user preferences for recommending movies
    • H04N21/4668—Learning process for intelligent management, e.g. learning user preferences for recommending movies for recommending content, e.g. movies

Landscapes

  • Engineering & Computer Science (AREA)
  • Databases & Information Systems (AREA)
  • Multimedia (AREA)
  • Signal Processing (AREA)
  • Theoretical Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Computer Graphics (AREA)
  • Computing Systems (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

The embodiment of the application discloses a data recommendation method, electronic equipment and a storage medium, which are applied to the technical field of machine learning. The method comprises the steps of obtaining a recommendation information set, generating a recommendation feature set according to the recommendation information set, determining a first recommendation index of a target object for a resource to be recommended according to the recommendation feature set, determining a second recommendation index of the target object for the resource to be recommended under the condition of target scene information according to the object attribute features and the resource attribute features, determining a target recommendation index of the target object for the resource to be recommended according to the first recommendation index and the second recommendation index, and pushing the resource to be recommended to an object terminal of the target object according to the target recommendation index. By adopting the embodiment of the application, the recommended prediction efficiency and accuracy of the resource to be predicted can be improved.

Description

Data recommendation method, electronic device and storage medium
Technical Field
The present application relates to the field of machine learning technologies, and in particular, to a data recommendation method, an electronic device, and a storage medium.
Background
The aim of the current recommendation task is mainly to predict resources to be recommended which are interested by a user in a target application and push the resources to be recommended to a user terminal, so that the effective recommendation rate for the user is improved. For example, a host of interest to a user is predicted and recommended to a live room of the host, thereby improving the click rate of the user on the live room. The existing prediction mode is usually to construct a prediction model to predict the relevant attributes of the user and the relevant attributes of the resources to be predicted so as to realize accurate recommendation. However, in this way, the model learns fewer features only by the two aforementioned attributes, resulting in low recommended prediction efficiency and accuracy. Therefore, how to improve the prediction efficiency and accuracy of the recommendation of the resource to be recommended becomes a problem to be solved.
Disclosure of Invention
The embodiment of the application provides a data recommendation method, electronic equipment and a storage medium, which can improve the recommendation prediction efficiency and accuracy of resources to be recommended under a multi-recommendation scene.
In one aspect, an embodiment of the present application provides a data recommendation method, where the method includes:
the method comprises the steps of acquiring a recommendation information set, wherein the recommendation information set comprises object attribute information of a target object, resource attribute information of a resource to be recommended and recommendation scene information of the resource to be recommended;
generating a recommendation feature set according to the recommendation information set, wherein the recommendation feature set comprises object attribute features of object attribute information, resource attribute features of resource attribute information and recommendation scene features of recommendation scene information;
determining a first recommendation index of the target object for the resource to be recommended according to the recommendation feature set, wherein the first recommendation index represents the probability that the target object responds to the recommendation behavior of the resource to be recommended;
determining a second recommendation index of the target object for the resource to be recommended under the target scene information according to the object attribute characteristics and the resource attribute characteristics, wherein the target scene information is at least one scene information in a plurality of scene information associated with the resource to be recommended, and the second recommendation index characterizes the probability that the target object responds to the recommendation behavior of the resource to be recommended;
determining a target recommendation index of a target object for the resource to be recommended according to the first recommendation index and the second recommendation index, pushing the resource to be recommended to an object terminal of the target object according to the target recommendation index, and representing target probability of response recommendation behaviors of the target object to the resource to be recommended.
In one aspect, an embodiment of the present application provides a data recommendation device, including:
The system comprises an acquisition module, a recommendation information collection and a recommendation information collection, wherein the acquisition module is used for acquiring a recommendation information collection, and the recommendation information collection comprises object attribute information of a target object, resource attribute information of a resource to be recommended and recommendation scene information of the resource to be recommended;
The processing module is used for generating a recommendation feature set according to the recommendation information set, wherein the recommendation feature set comprises object attribute features of object attribute information, resource attribute features of resource attribute information and recommendation scene features of recommendation scene information;
The determining module is used for determining a first recommendation index of the target object for the resource to be recommended according to the recommendation characteristic set, wherein the first recommendation index represents the probability that the target object responds to the recommendation behavior of the resource to be recommended;
the determining module is used for determining a second recommendation index of the target object for the resource to be recommended under the target scene information according to the object attribute characteristics and the resource attribute characteristics, wherein the target scene information is at least one scene information in a plurality of scene information associated with the resource to be recommended, and the second recommendation index represents the probability that the target object responds to the recommendation behavior of the resource to be recommended;
The determining module is used for determining a target recommendation index of a target object for the resource to be recommended according to the first recommendation index and the second recommendation index, wherein the target recommendation index represents target probability of response recommendation behavior of the target object for the resource to be recommended;
And the processing module is used for pushing the resources to be recommended to the object terminal of the target object according to the target recommendation index.
In one aspect, an embodiment of the present application provides an electronic device, including a processor and a memory, where the memory is configured to store a computer program, the computer program including program instructions, and the processor is configured to invoke the program instructions to perform some or all of the steps in the above method.
In one aspect, embodiments of the present application provide a computer readable storage medium storing a computer program comprising program instructions for performing part or all of the steps of the above method when executed by a processor.
Accordingly, according to one aspect of the present application, there is provided a computer program product or computer program comprising program instructions stored in a computer readable storage medium. The processor of the computer device reads the program instructions from the computer-readable storage medium, and the processor executes the program instructions, so that the computer device performs the data recommendation method provided above.
According to the method and the device for recommending the resources, a recommendation information set can be obtained, a recommendation feature set can be generated, a first recommendation index of a target object for the resources to be recommended is determined according to the recommendation feature set, the first recommendation index reflects interaction of various recommendation features in the recommendation feature set, a second recommendation index of the target object under target scene information is determined according to object attribute features and resource attribute features, the second recommendation index is obtained through learning features under the target scene information, the target recommendation index is determined according to the first recommendation index and the second recommendation index, the resources to be recommended are pushed to an object terminal of the target object according to the target recommendation index, the obtained target recommendation index can be combined with various recommendation features of the target object in a deeper level, and can be combined with learning features under various recommendation scenes, so that recommendation prediction efficiency and accuracy of the resources to be predicted can be improved, and the effect of follow-up recommendation based on the target recommendation index can be improved.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for the description of the embodiments will be briefly described below, and it is obvious that the drawings in the following description are some embodiments of the present application, and other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a schematic diagram of an application architecture according to an embodiment of the present application;
fig. 2 is a schematic flow chart of a data recommendation method according to an embodiment of the present application;
FIG. 3a is a schematic diagram of a recommendation scenario information provided in an embodiment of the present application;
FIG. 3b is a schematic diagram of a recommendation scenario information provided in an embodiment of the present application;
FIG. 3c is a schematic diagram of a recommendation scenario information provided in an embodiment of the present application;
fig. 4 is a flow chart of a data recommendation method according to an embodiment of the present application;
FIG. 5 is a schematic view of a scenario for generating a recommendation feature set according to an embodiment of the present application;
FIG. 6a is a schematic view of a scenario for determining fusion attribute characteristics according to an embodiment of the present application;
FIG. 6b is a schematic view of a scenario for determining fusion attribute characteristics according to an embodiment of the present application;
Fig. 7 is a schematic diagram of a pushing scenario of a resource to be recommended based on a target prediction model according to an embodiment of the present application;
fig. 8 is a flow chart of a data recommendation method according to an embodiment of the present application;
FIG. 9a is a schematic diagram of a scenario for model training according to an embodiment of the present application;
FIG. 9b is a schematic diagram of a prediction model according to an embodiment of the present application;
Fig. 10 is a schematic structural diagram of a data recommendation device according to an embodiment of the present application;
Fig. 11 is a schematic structural diagram of an electronic device according to an embodiment of the present application.
Detailed Description
The following description of the embodiments of the present application will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present application, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the application without making any inventive effort, are intended to be within the scope of the application.
The data recommendation method provided by the embodiment of the application is implemented in the electronic equipment, and the electronic equipment can be a server or a terminal. The server may be an independent physical server, a server cluster or a distributed system formed by a plurality of physical servers, or a cloud server providing cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDNs (Content Delivery Network, content delivery networks), basic cloud computing services such as big data and artificial intelligent platforms, and the like. The terminal may be, but is not limited to, a smart phone, a tablet computer, a notebook computer, a desktop computer, a smart speaker, a smart watch smart voice interaction device, a smart home appliance, a vehicle-mounted terminal, an aircraft, and the like.
In some embodiments, please refer to fig. 1, fig. 1 is a schematic diagram of an application architecture according to an embodiment of the present application, through which the data recommendation method according to the present application can be executed. The method comprises the steps of enabling a target object to be a target application, enabling the target object to be a user (such as a user) facing the target application to be a target application, enabling the electronic device to obtain a recommendation information set according to the target object and the target object to be recommended, generating a recommendation feature set, enabling the recommendation feature set to comprise a first recommendation feature of the target object, a second recommendation feature of the target object and a third recommendation feature of a recommendation scene where the target object is located, obtaining a first recommendation index of the target object for the target object to be recommended according to the three recommendation features, determining a second recommendation index according to the first recommendation feature and the second recommendation feature and combining the recommendation scene, determining a target recommendation index according to the first recommendation index and the second recommendation index, determining a recommendation strategy of the target object to be recommended in the target terminal according to the target recommendation index, and pushing the target object to the target terminal. The number of resources to be recommended may be one or more, for example only. Optionally, the above process may be further performed by a target prediction model, that is, a target prediction model may be deployed in the electronic device, a recommendation feature set is generated in the target prediction model through a recommendation information set, and a first recommendation index, a second recommendation index, and a target recommendation index are obtained according to the recommendation feature set.
It should be understood that fig. 1 is merely exemplary to represent possible application architectures of the technical solution of the present application, and is not limited to specific architectures of the technical solution of the present application, that is, the technical solution of the present application may also provide other application architectures.
Optionally, in some embodiments, the electronic device may execute the data recommendation method according to actual service requirements to improve the prediction efficiency and accuracy of the resource to be predicted. The electronic device can acquire corresponding recommendation information sets according to the target object and the object to be recommended to generate a recommendation feature set, wherein the recommendation feature set not only comprises relevant attribute features of the target object and relevant attribute features of resources to be predicted, but also comprises recommendation scene features corresponding to recommendation scene information where the resources to be predicted are located, and determines a first recommendation index and a second recommendation index according to the recommendation feature set and a recommendation scene where the combined object to be recommended is located, so that the target recommendation index is determined, the object to be recommended can be pushed to an object terminal of the target object according to the target recommendation index, and therefore accurate pushing and improving of recommendation effects can be achieved.
The recommended scene information may be any one of scene information associated with the resource to be recommended. For example, the scene information may refer to a scene of recommending the resource to be recommended in an application home page of the target application to which the resource to be recommended belongs, and for example, the scene information may also refer to a scene of recommending the resource to be recommended in a presentation page of other resources, and so on.
Optionally, the technical scheme of the application can be applied to any recommended task. For example, the method and the device can be applied to a live broadcast recommendation task, the object to be recommended at the moment can be a current online anchor, the target object can be a user using live broadcast application, the electronic equipment can determine the target recommendation index of the user for the anchor based on the technical scheme of the application, and push the anchor to the user terminal of the user based on the target recommendation index, so that the click rate, the watching duration and the like of the user for the recommended anchor are improved. For example, the method and the device can be applied to an e-commerce recommendation task, at the moment, the object to be recommended can be a commodity, the target object can be a user using the e-commerce application, the electronic equipment can determine the target recommendation index of the user for the commodity based on the technical scheme of the application, and the commodity is pushed to the user terminal of the user based on the target recommendation index, so that the click rate, the purchase rate and the like of the user for the recommended commodity are improved.
Optionally, the data related to the present application, such as the recommended information set, the recommended feature set, etc., may be stored in a database, or may be stored in a blockchain, such as by a blockchain distributed system, which is not limited by the present application.
It can be understood that the above scenario is merely an example, and does not constitute a limitation on the application scenario of the technical solution provided by the embodiment of the present application, and the technical solution of the present application may also be applied to other scenarios. For example, as one of ordinary skill in the art can know, with the evolution of the system architecture and the appearance of new service scenarios, the technical solution provided by the embodiment of the present application is also applicable to similar technical problems.
Based on the above description, the embodiments of the present application provide a data recommendation method, which may be performed by the above-mentioned electronic device. Referring to fig. 2, fig. 2 is a flowchart of a data recommendation method according to an embodiment of the present application. As shown in fig. 2, the flow of the data recommendation method according to the embodiment of the present application may include the following steps:
S201, acquiring a recommendation information set.
The recommendation information set may include object attribute information of a target object, resource attribute information of a resource to be recommended, and recommendation scene information of the resource to be recommended.
In some embodiments, the target object may be a user, the object attribute information may include one or more user attributes usable to characterize the relevant characteristics of the user, and the resource attribute information of the resource to be recommended may include one or more resource attributes usable to characterize the relevant characteristics of the resource to be recommended. The specific attribute types contained in the object attribute information and the resource attribute information can be set by related service personnel according to the recommended task of the actual application, and the specific attribute types are not limited herein. For example, taking a live broadcast recommendation task as an example, object attribute information of a target object may include basic portrait attributes (such as age, gender, academic, etc. of a user), statistical attributes (such as total time of live broadcast viewing in one or three days of the user, total flower number of a live broadcast in three or one week of the user, total number of live broadcast views in three or one week of the user, etc.), etc., resources to be recommended may be online live broadcast to be recommended, so resource attribute information of resources to be recommended may include basic portrait attributes (such as age, gender, academic, etc. of the live broadcast), real-time attributes (such as total number of viewing users or highest number of viewing users of live broadcast of the live broadcast in one or three days of the live broadcast, live broadcast time of the live broadcast, etc.), etc.
It will be appreciated that in the specific embodiments of the present application, related data of user information such as age of a user is involved, and when the above embodiments of the present application are applied to specific products or technologies, user permission or consent is required, and the collection, use and processing of related data is required to comply with related laws and regulations and standards of related countries and regions.
In some embodiments, the resource to be recommended belongs to a target application in the target terminal, the resource to be recommended can be recommended in multiple application interfaces in the target application, the multiple application interfaces can be regarded as multiple scene information which can be recommended and is associated with the resource to be recommended in the target application, and the resource to be recommended can be displayed in different presentation forms in the multiple application interfaces when the resource to be recommended is recommended. The plurality of scene information associated with the resource to be recommended is related to an application to which the resource to be recommended belongs.
Therefore, the recommended scene information of the resource to be recommended is any one of a plurality of scene information associated with the resource to be recommended, the process of executing the recommended prediction of the resource to be recommended under different recommended scene information is the same, and the prediction resources (such as related parameters participating in the process and the like) used by the different recommended scene information in the recommended prediction can be partially shared and partially exclusive, so that the technical scheme of the application can realize the recommended prediction under multiple recommended scenes. For example, the recommended scene information may be an application interface currently displayed in the target application, that is, scene information that can be recommended and in which the resource to be recommended is currently located. The electronic device may be a background device to which the target application belongs, when the object terminal detects that the target application displays a target application interface capable of being recommended, a recommendation instruction is generated and sent to the electronic device, and when the electronic device receives the recommendation instruction, the electronic device may acquire object attribute information of the target object and resource attribute information of a resource to be recommended according to the instruction of the recommendation instruction, and determine scene information represented by the target application interface as recommended scene information of the resource to be recommended.
The electronic device may construct a recommendation information set for each resource to be recommended when the plurality of the obtained resources to be recommended are available, where each recommendation information set includes object attribute information of the target object, resource attribute information of the resource to be recommended, and current recommendation scene information. The prediction process and principle of each recommended information set are the same, and the recommended information set is taken as an example for explanation.
For example, as shown in fig. 3 a-3 b, fig. 3 a-3 b are schematic diagrams of recommended scene information provided in the embodiment of the present application, where, taking a live broadcast recommendation task as an example, a resource to be recommended is a live broadcast in a live broadcast application, and may be a recommendation of a live broadcast in a live broadcast channel page (i.e. scene information 1) of the live broadcast application, where a live broadcast room where the live broadcast is located is denoted as a live broadcast tag (tab), and thumbnail displays are performed on the live broadcast channel page, as shown in fig. 3 a;
As another example, the live broadcast application may also recommend the live broadcast on a recommended channel page (i.e. scene information 2), where the details of the live broadcast room of the live broadcast are directly displayed on the recommended channel page, and the user may enter the live broadcast rooms of different recommended live broadcast by sliding up and down, as shown in fig. 3b, where one sliding indicates that one live broadcast is recommended;
As another example, the live broadcast application may also recommend the live broadcast on the live broadcast watching page (i.e. the scene information 3), the target application may display the live broadcast room detail picture of the live broadcast a in response to the triggering operation of the user on the live broadcast 1, may recommend the live broadcast information stream (feed) on the detail interface, that is, recommend other live broadcast on the designated area of the detail picture, as shown in fig. 3c, may display the live broadcast room thumbnail of the live broadcast 2 in the form of a floating layer of a floating window in the lower left corner, where the live broadcast room thumbnail may be the same as or different from the thumbnail in fig. 3a, and at this time, the recommendation mode may be that the live broadcast is recommended once in the designated time, such as recommending 10 live broadcast at most once every 10 s.
It can be understood that the technical scheme of the application can be applied to any recommended task, and the target object, the resource to be recommended and the recommended scene information can be correspondingly different according to different recommended tasks, and the recommended tasks are not limited. For convenience of explanation, the live recommendation task will be described hereinafter as an example.
S202, generating a recommendation characteristic set according to the recommendation information set.
The recommendation feature set may include an object attribute feature of the object attribute information, a resource attribute feature of the resource attribute information, and a recommendation scene feature of the recommendation scene information.
In one possible embodiment, the generation process and principle of each recommendation feature in the recommendation feature set are the same, and object attribute features are taken as an example. The generating, by the electronic device, the object attribute feature according to the object attribute information may specifically be performing One-Hot encoding on the object attribute information to obtain an encoded feature vector, and using the encoded feature vector as the object attribute feature. For example, the object attribute information includes an age attribute, and the age attribute is classified into a plurality of attribute categories of [ <18,19-30,31-40,41-50,51-60, >60], and if the age of the object is 24, the classified target attribute category is [19-30], and thus the encoded feature vector corresponding to the age attribute obtained by performing the thermal encoding may be represented as [0,1,0,0,0,0].
Corresponding encoded feature vectors may thus be generated from the one or more attributes contained in the object attribute information, respectively, from which the object attribute features may be composed. The partitioning method for the attribute can be set by relevant service personnel according to experience values when the hot independent coding is carried out. The specific manner in which the electronic device generates the resource attribute features and the recommended scene features may be the same as the specific manner in which the object attribute features are generated, and will not be described in detail herein.
In One possible embodiment, the electronic device may specifically generate the object attribute feature according to the object attribute information, perform One-Hot encoding (One-Hot encoding) on the object attribute information to obtain an encoded feature vector, obtain an embedded feature vector corresponding to the encoded feature vector, and use the embedded feature vector as the object attribute feature. The electronic device obtains an embedded feature vector corresponding to the coding feature vector, which may be an embedded feature matrix constructed for the object attribute information, and determines the corresponding embedded feature vector from the embedded feature matrix according to the coding feature vector and in a preset manner.
In some embodiments, the electronic device may specifically determine, according to the encoded feature vector and in a preset manner, a corresponding embedded feature vector from the embedded feature matrix, where the encoded feature vector is multiplied by the embedded feature matrix, and the product result is used as the corresponding embedded feature vector. Since the coding feature vector is composed of 0 and 1, the obtained embedded feature vector represents that the column value of the element with 1 in the coding feature vector corresponds to the target row vector of the embedded feature matrix. Corresponding embedded feature vectors may thus be generated from the one or more attributes contained in the object attribute information, respectively, from which object attribute features may be composed.
For example, since the object attribute information includes an age attribute, and the encoded feature vector obtained by performing the thermal unique encoding based on the age attribute is [0,1,0,0,0,0] and the column value of the element 1 in the encoded feature vector is 2, the vector of the 2 nd row is obtained from the embedded feature matrix as the embedded feature vector corresponding to the age attribute. The embedded feature matrix constructed during determining the embedded feature vector can be set by related business personnel according to experience values, the embedded feature matrices corresponding to different attribute information can be the same or different, and the matrix size of the embedded feature matrix is not limited. The specific manner in which the electronic device generates the resource attribute features and the recommended scene features may be the same as the specific manner in which the object attribute features are generated, and will not be described in detail herein.
Optionally, the object attribute feature, the resource attribute feature and the recommendation scene feature in the recommendation information set may all include a coding feature vector and an embedding feature vector that respectively correspond to each other.
S203, determining a first recommendation index of the target object for the resource to be recommended according to the recommendation characteristic set.
The first recommendation index may characterize a probability that the target object responds to a recommendation behavior for the resource to be recommended.
In some embodiments, when the target object is interested in the recommended resource to be recommended, the recommended behavior may be responded to the resource to be recommended, where the recommended behavior may be understood as that the resource to be recommended makes an effective recommendation for the target object, and the first recommendation index may also represent the interest degree of the target object in the resource to be recommended. Therefore, when the indicated value of the first recommendation index is larger, the probability of the target object responding to the recommendation behavior is larger, the probability of the to-be-recommended resource generating effective recommendation is larger, and the interest degree of the target object to the to-be-recommended resource is also larger. The specific type of the recommended behavior is related to the recommended scene information of the resource to be recommended, the recommended scene information is different, the recommended behavior may be different, and the type of the recommended behavior is not limited herein.
For example, taking a live broadcast recommendation task as an example, if the recommendation scenario information is the scenario information indicated in fig. 3a, the recommendation behavior may be a click behavior of a live broadcast room in a recommended channel page, where the first recommendation index may represent a predicted click rate of a target object on the recommended live broadcast, if the recommendation scenario information is the scenario information indicated in fig. 3b, the recommendation behavior may be a viewing behavior of a live broadcast room in the recommended channel page, where if the live broadcast room for viewing the live broadcast a is full of 3s, the live broadcast a is indicated as generating an effective recommendation, where the first recommendation index may represent a predicted viewing rate of the target object on the recommended live broadcast, and if the recommendation scenario information is the scenario information indicated in fig. 3c, the recommendation behavior may be a click behavior of other live broadcast on the recommended live broadcast page, where the first recommendation index may represent a predicted click rate of the target object on the recommended live broadcast.
In some embodiments, the electronic device may obtain the first recommendation index according to the plurality of independent recommendation features and the fused recommendation features in the recommendation feature set, so that the first recommendation index includes not only direct display feature information obtained based on the plurality of independent recommendation features, but also feature information implicitly displayed based on the plurality of fused recommendation features. The specific manner of obtaining the first recommendation index according to the plurality of independent recommendation features in the recommendation feature set and the recommendation features subjected to fusion together can be referred to the related description of the following embodiments.
In some embodiments, the determining, by the electronic device, a first recommendation index of the target object for the resource to be recommended according to the recommendation feature set may be obtained by a target prediction model. The method specifically comprises the steps of obtaining a target prediction model, and predicting in the target prediction model according to object attribute characteristics, resource attribute characteristics and recommended scene characteristics to obtain a first recommended index, wherein the object attribute characteristics, the resource attribute characteristics and the recommended scene characteristics at the moment all comprise the coding feature vectors and the embedding feature vectors which are respectively corresponding to each other. When the plurality of recommended feature sets are provided, model parameters used in predicting the plurality of recommended feature sets in the target prediction model can be the same, namely when the prediction obtains a first recommendation index, each recommended feature set shares a part of model network of the target prediction model so as to share model information. The target prediction model is trained based on the process described in steps S801 to S804 in the following embodiments.
S204, determining a second recommendation index of the target object for the resource to be recommended under the target scene information according to the object attribute characteristics and the resource attribute characteristics.
The target scene information is at least one scene information of a plurality of scene information associated with the resource to be recommended, and the second recommendation index can represent probability that the target object responds to the recommended behavior associated with the target scene information.
In some embodiments, the second recommendation index may represent the interest level of the target object to the recommended resource under the target scene information, so when the indicated value of the second recommendation index is larger, the probability that the target object responds to the recommended action under the target scene information is larger, the probability that the resource to be recommended generates effective recommendation under the target scene information is also larger, and the interest level of the target object to the recommended resource under the target scene information is also larger.
In one possible implementation manner, the recommended scenario information is scenario information in which the resource to be recommended is currently located in the target application, and the plurality of scenario information associated with the resource to be recommended includes the recommended scenario information, and the target scenario information may be determined based on the recommended scenario information.
In some embodiments, the target scene information may be recommended scene information, and may also be recommended scene information and associated scene information of the recommended scene information among the plurality of scene information. The associated scene information of the recommended scene information may be all scene information except the recommended scene information in the plurality of scene information, or may be specified scene information in the plurality of scene information, which may be set by the relevant service personnel according to an experience value.
For example, let the recommended scene information be scene information 2 as indicated in fig. 3b, at this time, the recommended scene information be an application interface for playing the live-room picture, and the application interface represented by scene information 3 as indicated in fig. 3c also plays the live-room picture, so that the scene information 3 may be set as the associated scene information of the scene information 2.
In some embodiments, the second recommendation index may be derived from the target prediction model described above, which may include a plurality of scene prediction networks. Wherein one scene prediction network corresponds to one scene information, i.e. each scene information has a unique scene prediction network.
When the target scene information is recommended scene information, the electronic device determines, according to the object attribute feature and the resource attribute feature, a second recommendation index of the target object for the resource to be recommended under the target scene information, specifically, may be that the object attribute feature and the resource attribute feature are predicted according to a scene prediction network corresponding to the recommended scene information, so as to obtain the second recommendation index. Wherein, the object attribute feature and the resource attribute feature which are predicted are corresponding embedded feature vectors.
When the target scene information is the recommended scene information and the associated scene information, the electronic device determines, according to the object attribute feature and the resource attribute feature, a second recommendation index of the target object for the resource to be recommended under the target scene information, specifically, may predict the object attribute feature and the resource attribute feature according to a scene prediction network corresponding to the recommended scene information and a scene prediction network corresponding to the associated scene information, respectively, to obtain the second recommendation index. The specific manner of obtaining the second recommendation index according to the scene prediction network corresponding to the recommendation scene information and the scene prediction network corresponding to the associated scene information may be referred to the related description of the following embodiments. Wherein, the object attribute feature and the resource attribute feature which are predicted are corresponding embedded feature vectors.
It can be understood that the above-mentioned scene prediction network is trained by using object attribute features and resource attribute features under the respective corresponding scene information, and thus includes learning features unique to the respective corresponding scene information. When the prediction is performed, the object attribute features and the resource attribute features can be divided into corresponding exclusive scene prediction networks based on the target scene information to obtain a second recommendation index, wherein the second recommendation index comprises feature information of the object attribute features and the resource attribute features under the premise of learning features of the target scene information. When the target prediction model predicts different recommendation feature sets, the recommendation scene information in the recommendation feature sets is different, and the scene prediction network used is also different, that is, the model parameters used in the prediction can be different and different from the model parameters used in the first recommendation index. The first recommendation index and the second recommendation index are both obtained based on the recommendation feature set, but are obtained by different model parameters under different prediction angles. Based on the above description, the target prediction model may include at least a common prediction network for predicting the first recommendation index and a plurality of scene prediction networks for predicting the second recommendation index.
S205, determining a target recommendation index of a target object for the resource to be recommended according to the first recommendation index and the second recommendation index, and pushing the resource to be recommended to an object terminal of the target object according to the target recommendation index.
The target recommendation index can be obtained through the combination of the first recommendation index and the second recommendation index, and the prediction accuracy of the target recommendation index can be improved. The target recommendation index characterizes target probability that the target object responds to the recommended behavior of the resource to be recommended.
In some embodiments, the determining, by the electronic device, the target recommendation index of the target object for the resource to be recommended according to the first recommendation index and the second recommendation index may specifically be that the first recommendation index and the second recommendation index are weighted and summed to obtain the target recommendation index. Wherein the coefficients of the weighted summation can be set by the relevant service personnel.
Alternatively, determining the target recommendation index according to the first recommendation index and the second recommendation index may also be obtained through a target prediction model, and the target recommendation index may be output through the predicted first recommendation index and second recommendation index in the target prediction model. For example, the weighting coefficients may be obtained based on training as model parameters in the target prediction model.
In some embodiments, the target recommendation index may represent the interest degree of the target object to the resource to be recommended under the current recommendation scene information, so that a recommendation policy of the resource to be recommended may be determined based on the target recommendation index, and the resource to be recommended may be pushed to the object terminal based on the recommendation policy. The recommendation policy may be set by the relevant business personnel based on empirical values. For example, when there are a plurality of resources to be recommended, the recommendation policy may be to push the resources to be recommended to the target terminal in order from the higher to the lower target recommendation indexes, for example, the recommendation policy may also be to push the resources to be recommended when the target recommendation index is greater than or equal to the preset index, and not push the resources to be recommended when the target recommendation index is less than the preset index, and so on.
According to the method and the device for recommending the resources, a recommendation information set can be obtained, a recommendation feature set is generated according to the recommendation information set, a first recommendation index of a target object for the resources to be recommended is determined according to the recommendation feature set, interaction of various recommendation features in the recommendation feature set is reflected by the first recommendation index, a second recommendation index of the target object for the resources to be recommended is determined according to the object attribute features and the resource attribute features, the second recommendation index is obtained by utilizing learning features in target scene information, the learning features in the target scene information can be obtained comprehensively through learning features of all scene information, the target recommendation index of the target object for the resources to be recommended is determined according to the first recommendation index and the second recommendation index, the resources to be recommended are pushed to an object terminal of the target object according to the target recommendation index, and the obtained target recommendation index can be combined with various recommendation features of the target object in a deeper layer, and can be combined with learning features in various recommendation scenes, so that recommendation prediction efficiency and accuracy of the resources to be predicted can be improved, and the effect of the target recommendation index is improved.
Referring to fig. 4, fig. 4 is a flowchart of a data recommendation method according to an embodiment of the present application, where the method may be performed by the above-mentioned electronic device. As shown in fig. 4, the flow of the data recommendation method in the embodiment of the present application may include the following:
s401, acquiring a recommendation information set. The specific implementation manner of step S401 may be referred to the related description of the above embodiments, which is not repeated herein.
S402, generating a recommendation characteristic set according to the recommendation information set.
The recommendation characteristic set comprises object attribute characteristics of the object attribute information, resource attribute characteristics of the resource attribute information and recommendation scene characteristics of the recommendation scene information.
In one possible implementation manner, the generation process and principle of each recommended feature in the recommended feature set are the same, and each recommended feature may include a coded feature vector and an embedded feature vector obtained by each recommended information in the recommended information set. The specific method for generating the recommended feature set may be referred to in the description of the above embodiment.
For example, as shown in fig. 5, fig. 5 is a schematic view of a scene for generating a recommended feature set according to an embodiment of the present application; the object attribute information includes object attribute 1, and the coding feature vector corresponding to the object attribute 1 is [0,1,0,0,0,0] (U1), so that the row vector of the 2 nd row can be obtained from the corresponding embedding matrix 1 as the embedding feature vector corresponding to the object attribute 1 (V1), the corresponding object attribute feature is obtained based on the object attribute information (including object attribute 1, object attribute 2,..and object attribute N) as the coding feature vector [ U1, U2,..and Un ] and the embedding feature vector [ V1, V2,..and Vn ], and the corresponding resource attribute feature is obtained based on the resource attribute information (including resource attribute n+1, resource attribute n+2,..and resource attribute N) as the coding feature vector [ un+1, un+2,..un ] and the embedding feature vector [ vn+1, vn+2 ], and the corresponding scene feature is obtained based on the recommended scene information (set m) as the coding feature vector [ Um ] and the embedding feature vector [ Vm ].
Alternatively, the above process may also be obtained by the target prediction model, i.e. the target prediction model may include a feature generation layer, and the feature generation layer generates a recommended feature set from the recommended information set, i.e. the embedded feature matrix may be used as model parameters in the target prediction model. The target prediction model is trained based on the process described in steps S801 to S804 in the following embodiments.
S403, determining fusion attribute characteristics and recommendation influence values according to the recommendation characteristic set, and determining a first recommendation index according to the fusion attribute characteristics and the recommendation influence values.
In one possible implementation manner, the electronic device may determine the recommendation influence value according to multiple independent recommendation features in the recommendation feature set, and perform feature fusion according to the multiple independent recommendation features to obtain a fusion attribute feature, where the fusion attribute feature may deepen feature interactions between the multiple recommendation features to obtain more feature information.
In one possible implementation, the recommendation impact value may characterize how much a plurality of recommendation features in the recommendation information set impact the target object's response to the recommendation behavior. The determining of the recommendation impact value by the electronic device according to the recommendation feature set may be implemented by the target prediction model. The target prediction model may further include a public prediction network, the public prediction network may include an influence prediction network, the electronic device may obtain all the encoding feature vectors from the recommended feature set, and the influence prediction network predicts the recommended influence value according to all the encoding feature vectors. The coded feature vector is composed of 0 and 1, and only the element value corresponding to the target attribute category classified based on the attribute is 1, so that prediction of the recommended influence value can be performed by the coded feature vector, indicating prediction based on only the classified target attribute category.
Alternatively, the recommended influence value may be positive or negative, and the positive influence of the plurality of recommended features is greater when the recommended influence value is positive and greater, whereas the negative influence of the plurality of recommended features is greater when the recommended influence value is negative and greater. For example, if the recommendation influence value of the target object to the resource a to be recommended is determined to be 1 based on the recommendation information set 1 and the recommendation influence value of the target object to the resource B to be recommended is determined to be 2 based on the recommendation information set 2, the target object can respond to the recommendation behavior of the resource B to be recommended more easily under the influence of the recommendation information set 2.
In one possible implementation, the fusion attribute feature represents that a new fusion recommendation feature is generated by a plurality of recommendation features of a recommendation feature set, and the most differential information can be obtained from a plurality of original recommendation features related in the fusion process, and redundant information generated by correlation among different recommendation features can be eliminated. The determining of the fusion attribute feature by the electronic device according to the recommended feature set may be implemented by the target prediction model. The target prediction model may include a public prediction network, which may include a feature fusion network.
In some embodiments, the electronic device uses a feature fusion network to perform feature fusion according to the recommended feature set, so as to obtain the fusion attribute feature specifically, may be to obtain all embedded feature vectors in the recommended feature set, and perform feature fusion on all the embedded feature vectors.
In some embodiments, the number T of all embedded feature vectors (set to V) includes a first embedded feature vector (set to Vi) and a second embedded feature vector (set to Vj), so feature fusion of all embedded feature vectors may specifically be according to the following fusion formula:
P ij＝Vi*W*Vj T*wij formula 1.1;
The method comprises the steps of setting a common weight matrix W, setting W ij as a unique weight parameter for a first embedded feature vector and a second embedded feature vector, setting V j T as a transposed vector of the second embedded feature vector, setting i and j as positive integers from 1 to T, setting i as a default parameter set by related business personnel when i is equal to j, and setting P ij as a default parameter set by related business personnel when i is equal to j.
The size of the common weight matrix is nxm, and as the size of the embedded feature matrix corresponding to each attribute can be different, the obtained embedded feature vector elements can be different, a plurality of common weight matrices can be used when feature fusion is carried out, and the size of each common weight matrix is different and is determined based on the two fused embedded feature matrices. When the first embedded feature vector and the second embedded feature vector are fused, a matched common weight matrix can be selected from a plurality of common weight matrices based on the number of elements of the first embedded feature vector and the second embedded feature vector respectively;
alternatively, there may be only one target weight matrix, where the size of the target weight matrix is determined based on the number of elements of the target embedded feature vector, where the target embedded feature vector is the feature vector with the largest number of elements in all the embedded feature vectors, and if the number of elements is Nmax, the size of the target weight matrix may be a matrix of Nmax rows and Nmax columns, and when the first embedded feature vector and the second embedded feature vector are fused, n×m weight values are sequentially extracted from the top left corner of the matrix in the target weight matrix to form a required common weight matrix.
Based on the above fusion formula, one fusion element can be obtained for every two embedded feature vectors, a plurality of fusion elements can be obtained through all the embedded feature vectors, and a feature matrix formed by the plurality of fusion elements in sequence is used as the fusion feature matrix, wherein the fusion formula refers to the idea of WKFM (a weighted FM (Factorization Machine, factorizer)). Optionally, the common weight matrix and the unique weight parameters may be set by related service personnel according to experience values, or may be network parameters in the feature fusion network, and are obtained by training the target prediction model.
For example, as shown in fig. 6 a-6 b, fig. 6 a-6 b are schematic views of a scene for determining fusion attribute features according to an embodiment of the present application, where the recommended feature set includes an embedded feature vector 1 (set to V1, the number of elements is 3), an embedded feature vector 2 (set to V2, the number of elements is 4), and an embedded feature vector 3 (set to V3, the number of elements is 5), so that:
As shown in fig. 6a, there are a plurality of common weight matrices, the size of which is determined based on the number of elements of each embedded feature vector, which may be 3x4, 3x5, & etc., respectively; when the embedded feature vector 1 and the embedded feature vector 3 are fused, the common weight matrix 1 with the size of 3x4 is selected from a plurality of common weight matrixes according to the number of elements of the embedded feature vector 1 and the embedded feature vector 2, and the common weight matrix 2 with the size of 3x5 is selected from the plurality of common weight matrixes according to the number of elements of the embedded feature vector 1 and the embedded feature vector 3;
As another example, as shown in fig. 6b, when there is a target weight matrix, the target weight matrix may be a 5x5 matrix, and when feature fusion is performed, when the embedded feature vector 1 and the embedded feature vector 2 are fused, 3x4 weight values are sequentially extracted from the target weight matrix according to the number of elements of the embedded feature vector 1 and the embedded feature vector 2 to form a common weight matrix 1, when the embedded feature vector 1 and the embedded feature vector 3 are fused, 3x5 weight values are sequentially extracted from the target weight matrix according to the number of elements of the embedded feature vector 1 and the embedded feature vector 3 to form the common weight matrix 2;
The embedded feature vector 1 is multiplied by the common weight matrix 1, the transpose vector of the embedded feature vector 2, the embedded feature vector 1 and the unique weight parameter (w 12) corresponding to the embedded feature vector 2 in sequence according to the above fusion formula to obtain a fusion element (set as P 12), the embedded feature vector 1 is multiplied by the common weight matrix 2, the transpose vector of the embedded feature vector 3, the embedded feature vector 1 and the unique weight parameter (w 13) corresponding to the embedded feature vector 3 in sequence according to the above fusion formula to obtain a fusion element (set as P 13), 9 fusion elements can be obtained based on V1, V2 and V3, and the fusion feature matrix of 3*3 is the fusion attribute feature.
In some embodiments, the electronic device may determine the first recommendation index in a predetermined manner based on the fused attribute features and the recommendation impact values. The preset determination mode can be set by relevant service personnel when constructing the target prediction model.
For example, the preset determining mode may be that the sum of the elements in the fusion attribute feature and the recommendation influence value are weighted and summed to obtain the first recommendation index, for example, the preset determining mode may be that the fusion attribute feature is input into the fusion feature prediction network to obtain the prediction value, the sum of the prediction value and the recommendation influence value is determined to be the first recommendation index, the target prediction model may further include the fusion feature prediction network, for example, the preset determining mode may be that the fusion attribute feature and the recommendation influence value are input into the index prediction network together to obtain the first recommendation index, and the target prediction model may further include the index prediction network, for example. The predetermined determination method is not limited herein.
S404, determining a second recommendation index of the target object for the resource to be recommended under the target scene information according to the object attribute characteristics and the resource attribute characteristics.
The target scene information is at least one scene information of a plurality of scene information associated with the resource to be recommended, and the plurality of scene information can comprise recommended scene information of the resource to be recommended. The second recommendation index characterizes the probability that the target object responds to the recommended behavior of the resource to be recommended.
In one possible implementation, the determining, by the electronic device, the second recommendation index of the target scene information according to the object attribute feature and the resource attribute feature may be obtained from the target prediction model, where the target prediction model may include a scene prediction network for a plurality of scene information, and one scene prediction network corresponds to one scene information.
In some embodiments, the target scene information may be recommended scene information, and the electronic device may acquire a plurality of scene prediction networks in the target prediction model, and predict an embedded feature vector that is an object attribute feature and an embedded feature vector that is a resource attribute feature according to a scene prediction network corresponding to the recommended scene information in the plurality of scene prediction networks to obtain the second recommendation index.
In some embodiments, the target scene information may include recommended scene information and associated scene information of the recommended scene information of the plurality of scene information, the associated scene information may be one or more, and may be set by the associated business person according to an experience value. The electronic equipment can acquire a plurality of scene prediction networks in the target prediction model, predict object attribute characteristics and resource attribute characteristics according to scene prediction networks corresponding to recommended scene information in the scene prediction networks and scene prediction networks corresponding to associated scene information respectively to obtain a first initial recommendation index of a target object in the recommended scene information aiming at resources to be recommended and a second initial recommendation index of the target object in the associated scene information aiming at the resources to be recommended, and when the associated scene information is a plurality of, the corresponding second initial recommendation indexes are also a plurality of, and obtain a second recommendation index according to the first initial recommendation index and the second initial recommendation index.
In some embodiments, the electronic device may specifically obtain the second recommendation index according to the first initial recommendation index and the second initial recommendation index, where the weighted summation is performed on the first initial recommendation index and the second initial recommendation index to obtain the second recommendation index. The weighting coefficients can be set by relevant business personnel according to experience values.
And S405, determining a target recommendation index of the target object for the resource to be recommended according to the first recommendation index and the second recommendation index. The specific implementation manner of step S405 may be referred to the related description of the above embodiments, which is not repeated herein.
S406, pushing the resource to be recommended to the object terminal of the target object according to the target recommendation index.
In some embodiments, the electronic device may determine, based on the target recommendation index, a resource to be recommended that is most likely to be of interest to the target object, thereby increasing the effective recommendation rate. The pushing the resource to be recommended to the object terminal of the target object according to the target recommendation index may be pushing the resource to be recommended with the target recommendation index being greater than the preset index, or may be sequentially pushing the resource to be recommended according to the order of the target recommendation index from the higher one to the lower one, or the like.
For example, fig. 7 is a schematic diagram of a pushing scenario of a resource to be recommended based on a target prediction model, which is applied to a live recommendation task, where:
When the target application iN the user terminal displays an application interface indicated by any one of the plurality of scene information, the electronic device acquires a plurality of currently online anchors (i 1, i2, i3, & gt, iN order to compose an information pair with each anchor and the currently located recommended scene information(s) respectively, forms a prediction pair ([ u, i1, s ], [ u, i2, s ], [ u, i3, s ], [ u, iN, s ]), extracts user attribute information of the target user and anchor attribute information of each anchor based on a plurality of prediction pair attribute storage platforms (such as a database) to compose the user attribute information and the anchor attribute information of each anchor, sequentially inputs each information pair into a target prediction model, generates a recommendation attribute feature (comprising the user attribute information, the anchor attribute feature of the corresponding anchor attribute information, the scene information) of each information pair iN the target prediction model, sequentially sorts the recommendation attribute feature pair according to a recommendation index of each recommendation attribute information pair, and sequentially sorts the recommendation attribute information pair according to a second recommendation index, and sequentially sorts the recommendation attribute information pair by the second recommendation index.
Recommending based on the ordered anchor may specifically be, as shown in fig. 3a, sequentially obtaining live broadcast room thumbnails of the ordered anchor (such as anchor 1, anchor 3, anchor 2, and..once again.), and sequentially displaying the live broadcast room thumbnails on a live broadcast channel page of a target application; as shown in fig. 3b, the live broadcasting room detail pictures of the ordered anchor (such as anchor 1, anchor 3, anchor 2 and the above-mentioned) are sequentially acquired, the live broadcasting room detail picture of the anchor 1 is displayed on a recommended channel page of the target application, the live broadcasting watching scene of the anchor 1 can be entered by clicking the live broadcasting room through touch control, and when the sliding operation of the target user is detected, the live broadcasting room detail picture of the next anchor 3 is sequentially displayed on the recommended channel page to indicate that the recommendation behavior is executed.
According to the method, a recommendation information set can be obtained, a recommendation feature set is generated according to the recommendation information set, fusion attribute features and recommendation influence values are determined according to the recommendation feature set, a first recommendation index is determined according to the fusion attribute features and the recommendation influence values, the first recommendation index is obtained based on the fusion attribute features and can reflect interaction of various recommendation features in the recommendation feature set, the first recommendation index is also obtained based on the recommendation influence values and can reflect influence degrees of various recommendation features on response behaviors, a second recommendation index of a target object aiming at a resource to be recommended under target scene information is determined according to object attribute features and resource attribute features, the second recommendation index is obtained by learning features under the target scene information, learning features under the target scene information can be obtained comprehensively through learning features of all scene information, the target object is determined according to the first recommendation index and the second recommendation index, the resource to be recommended to the target object terminal according to the target recommendation index, the obtained target recommendation index can be combined with various features of the target object terminal, prediction efficiency can be improved under various scenes, and the prediction efficiency can be improved under the multiple scenes based on the prediction results, and the prediction results can be predicted in the multiple scenes.
Referring to fig. 8, fig. 8 is a flowchart of a data recommendation method according to an embodiment of the present application, and the method may be performed by the above-mentioned electronic device. As shown in fig. 8, the flow of the data recommendation method in the embodiment of the present application may include the following:
S801, acquiring a plurality of sample information sets.
The sample information set comprises sample object attribute information of a sample object, sample resource attribute information of a sample recommended resource and sample scene information of the sample recommended resource. Sample objects, sample recommendation resources, and sample scene information in the multiple sample information sets may all be different.
In some embodiments, the target recommendation index described above may be generated by a target prediction model. The electronic device may perform batch training on the initial predictive model using the plurality of sample information sets as a batch (batch) of sample data, and obtain a final target predictive model based on multiple batches of training until convergence. The process and principle of each batch training are the same, and the kth training is taken as an example for illustration. Sample objects, sample recommended resources, and sample scene information in the multiple sample information sets of the kth round may all be different. The sample information set may include sample data as used in the application process described in the above embodiment.
S802, inputting each sample information set into an initial prediction model, and generating a sample feature set corresponding to each sample information set based on the initial prediction model.
The sample feature set comprises sample object attribute features of sample object attribute information, sample resource attribute features of sample resource attribute information and sample recommended scene features of sample scene information of the sample resource attribute information.
In some embodiments, the initial predictive model may include a feature generation layer to be trained, where model parameters may include an embedded feature matrix corresponding to each attribute, which may be derived from model training. The electronic device may sequentially input each sample information set into the initial prediction model, generate a coded feature vector and an embedded feature vector corresponding to each sample information set based on a feature generation layer in the initial prediction model, and use the corresponding coded feature vector and the embedded feature vector as a sample feature set corresponding to each sample information set. The specific manner of generating the encoded feature vector and the embedded feature vector corresponding to each sample information set based on the embedded feature matrix corresponding to each attribute in the feature generation layer may be the same as that described in the above embodiment.
S803, determining target sample recommendation indexes of sample recommendation resources corresponding to the sample objects according to the sample feature sets.
In some embodiments, the process and principle of determining, by the electronic device, the target sample recommendation index corresponding to each sample information set according to each sample feature set are the same, and here, description is given by taking any one sample feature set in each sample feature set as an example, where any one sample feature set is taken as a target feature set, a target sample object is a sample object corresponding to the target feature set, a target sample recommendation resource is a sample recommendation resource corresponding to the target sample object, and target sample scene information is sample scene information corresponding to the target sample recommendation resource.
The method comprises the steps of generating a first sample recommendation index of a target sample object for a target sample recommendation resource according to a target feature set in an initial prediction model, determining a second sample recommendation index of the target sample object for the target sample recommendation resource according to sample object attribute features and sample resource attribute features in the target feature set, and determining the target sample recommendation index of the target sample object for the target sample recommendation resource according to the first sample recommendation index and the second sample recommendation index.
The first sample recommendation index can represent initial sample probability of the target sample object to respond to the recommended action of the target sample recommended resource, the second sample recommendation index can represent initial sample probability of the target sample object to respond to the recommended action of the target sample recommended resource under the condition of target sample scene information, and the target sample recommendation index can represent target sample probability of the target sample object to respond to the recommended action of the target sample recommended resource.
In one possible embodiment, the initial prediction model may include an initial common prediction network and a plurality of initial scene prediction networks, one initial scene prediction network corresponding to each sample of scene information. The generating, by the electronic device, a first sample recommendation index of the target sample object for the target sample recommendation resource according to the target feature set in the initial prediction model may specifically be generating, by the electronic device, the first sample recommendation index according to the target feature set in the initial public prediction network. The determining, by the electronic device, a second sample recommendation index of the target sample object for the target sample recommended resource according to the sample object attribute feature and the sample resource attribute feature in the target feature set may specifically be determining, in the target scene prediction network, the second sample recommendation index according to an embedded feature vector of the sample object attribute feature and an embedded feature vector of the sample resource attribute feature in the target feature set, where the target scene prediction network is an initial scene prediction network corresponding to target sample scene information in a plurality of initial scene prediction networks.
In some embodiments, the initial public prediction network may include an initial feature fusion network and an initial influence prediction network, where the electronic device generates a first sample recommendation index according to the target feature set in the initial public prediction network, where feature fusion is performed on all embedded feature vectors in the target feature set to obtain sample fusion attribute features of the target information set, feature fusion is performed on all coded feature vectors in the target feature set in the initial influence prediction network to obtain sample recommendation influence values of the target information set, and sum all element values in the sample fusion attribute features to obtain sample fusion values, and a sum of the sample fusion values and the sample recommendation influence values is used as the first sample recommendation index.
S804, correcting model parameters of the initial prediction model according to the recommendation index of each target sample to obtain a target prediction model.
In some embodiments, the electronic device may specifically correct the model parameters of the initial prediction model according to each target sample recommendation index to obtain a target prediction model, correct the network parameters of the initial public prediction network according to each target sample recommendation index to obtain a public prediction network, correct the network parameters of a plurality of initial scene prediction networks according to each target sample recommendation index and sample scene information of each sample recommendation resource to obtain a plurality of scene prediction networks, and determine the target prediction model according to the public prediction network and the plurality of scene prediction networks. And obtaining a final target prediction model by correcting the model parameters for multiple times. If the model parameters of the initial prediction model further include an embedded feature matrix in the feature generation layer and a weighting coefficient for performing weighted summation, the model parameters of the foregoing part may be corrected by using a recommendation index of each target sample, and a final target prediction model may be obtained by using a common prediction network, a plurality of scene prediction networks, a trained embedded feature matrix, and a weighting coefficient.
In some embodiments, the electronic device corrects network parameters of the plurality of initial scene prediction networks according to the sample scene information of each target sample recommendation index and each sample recommendation resource, and the obtaining of the plurality of scene prediction networks may specifically be that each target sample recommendation index is respectively divided into the plurality of initial scene prediction networks according to the sample scene information of each sample recommendation resource, and the network parameters of the divided initial scene prediction networks are sequentially corrected according to each target sample recommendation index, so as to obtain the plurality of scene prediction networks in the target prediction model.
For example, the plurality of initial scene prediction networks include a prediction network A, a prediction network B and a prediction network C, the sample scene information A in the sample information set A corresponds to the prediction network A, the sample scene information B in the sample information set B corresponds to the prediction network B, the sample scene information C in the sample information set C corresponds to the prediction network C, the target sample recommendation index A corresponding to the sample information set A is obtained according to a first sample recommendation index A and a second sample recommendation index A generated based on the prediction network A, the target sample recommendation index B corresponding to the sample information set B is obtained according to the first sample recommendation index B and a second sample recommendation index B generated based on the prediction network B, and the target sample recommendation index C corresponding to the sample information set C is obtained according to the first sample recommendation index C and the second sample recommendation index C generated based on the prediction network C.
In some embodiments, the electronic device may obtain a recommendation index tag for each sample information set separately and determine a loss value for each sample information set based on the target sample recommendation index and the recommendation index tag for each sample information set. Thus modifying the network parameters of the initial public prediction network according to the recommendation index of each target sample may be modifying the network parameters of the initial public prediction network according to the loss value of each set of sample information. The correcting network parameters of the plurality of initial scene prediction networks according to the sample scene information of each target sample recommendation index and each sample recommendation resource may be respectively correcting network parameters of the initial scene prediction network corresponding to the sample scene information of each sample information set according to the loss value of each sample information set. And the electronic device may determine a target loss value for a plurality of sample information sets from the loss value for each sample information set and determine whether the trained model converges based on the target loss value.
For example, as shown in FIG. 9a, FIG. 9a is a schematic view of a model training scenario provided by an embodiment of the present application, an initial prediction model may include a feature generation layer, a first recommendation index generation layer, a second recommendation index generation layer, and a target recommendation index generation layer, the first recommendation index generation layer may include an initial public prediction network, the second recommendation index generation layer may include a plurality of initial scene prediction networks, each initial scene prediction network corresponds to one sample scenario information, the initial scene prediction network may be composed of one or more fully connected layers (Fully Connected Layers, FC), the number of fully connected layers of different initial public prediction networks may be different, and may be set by related business personnel according to experience values, the initial public prediction network may include an initial feature fusion network and an initial impact prediction network, the initial impact prediction network may be composed of one or more fully connected layers, wherein:
when a plurality of sample information sets are acquired, generating a sample feature set corresponding to each sample information set at a feature generation layer, wherein the sample feature set comprises a corresponding coding feature vector and an embedded feature vector;
respectively obtaining sample fusion attribute characteristics corresponding to each sample information set in an initial characteristic fusion network according to the embedded characteristic vector in each sample characteristic set;
Respectively obtaining sample recommendation influence values corresponding to each sample information set according to embedded feature vectors in each sample feature set in an initial influence prediction network, and determining the sum of all elements in sample fusion attribute features corresponding to each sample information set and the sample recommendation influence values corresponding to the respective sample information sets as a first sample recommendation index corresponding to each sample information set;
Predicting sample object attribute features and sample resource attribute features in each sample information set according to an initial scene prediction network corresponding to sample scene information in each sample information set to obtain a second sample recommendation index corresponding to each sample information set;
Respectively carrying out weighted summation on the first sample recommendation index and the second sample recommendation index corresponding to each sample information set in the target recommendation index generation layer to obtain a target sample recommendation index corresponding to each sample information set;
And in the application stage, if the target scene information is a plurality of, determining a weighting coefficient used for the second recommendation index to carry out weighted summation on the first initial recommendation index and the second initial recommendation index, wherein the weighting coefficient can be appointed in the target prediction model by related business personnel, or can be further trained based on the target prediction model, only the weighting coefficient of the part is corrected, so that the final target prediction model is obtained.
Optionally, a gating unit may be added to the initial prediction model to distinguish different initial scene prediction networks based on different scene information, and the gating unit may determine a corresponding target scene prediction network according to target sample scene information, transfer sample object attribute features and sample resource attribute features in a target feature set into the target scene prediction network, and hide initial scene prediction networks other than the target scene prediction network from the target feature set.
It will be appreciated that the model structure of the initial predictive model described above is merely exemplary and that other forms are possible. For example, the model structure constructed may be different based on the manner in which the first sample recommendation index, and/or the second sample recommendation index, and/or the target sample recommendation index are determined. If the initial public prediction network may further include an initial index prediction network, the initial index prediction network may be formed by one or more fully connected layers, and after obtaining the sample fusion attribute feature and the sample recommendation influence value, the first sample recommendation index may be determined based on the initial index prediction network, as shown in fig. 9b, where fig. 9b is a schematic diagram of a prediction model provided by an embodiment of the present application.
Through the above procedure, model prediction of multiple recommended scenes can be achieved, the resulting target prediction model may include a common network (common tower) for all recommended scenes and a unique network (unique tower) specific to a certain recommended scene, and the final output is determined by the output of the common tower and the output of the unique tower. The prediction resource of the target prediction model is shared by a plurality of recommended scenes of part of resources, and the part of resources are special for each recommended scene information, so that the learning features blended in the recommendation prediction are more comprehensive, and the prediction effect is better. During training, a plurality of sample scene information sample information sets can be included in one batch of samples, so that scene sub-tower training based on batch data is realized, one batch of data is not needed, only the same sample scene information is needed, namely, the sample data is subjected to sub-tower training in one batch of training process, so that samples aiming at different recommended scenes are segmented to train different scene prediction networks, and meanwhile, the samples of the different recommended scenes can be jointly trained into a public prediction network, so that feature differences among different recommended scenes can be fully learned, and features among different recommended scenes can be fully utilized and cooperated.
In addition, by jointly modeling a plurality of recommended scenes, the recommended targets of different recommended scenes can be aligned, and the recommended targets of different recommended scenes can be different, so that the recommended targets of each recommended scene can be equivalently mapped, the relevance of different recommended scenes is enhanced, and the model prediction effect is better. For example, the recommendation goal of the recommendation scenario shown in fig. 3a is to make the target object click on the resource to be recommended to generate an effective recommendation, and the recommendation goal of the recommendation scenario shown in fig. 3b is to make the target object watch the resource to be recommended to a specified duration to generate an effective recommendation.
For another example, taking a live broadcast recommendation task as an example, performing a large amount of measurement on the target prediction model obtained through training to find that the prediction accuracy and efficiency of the target prediction model are greatly improved compared with the prior art, and the click rate and the watching time length of a user on a live broadcast room are greatly improved through the recommendation mode;
See table 1 below:
data set Exposure to light Clicking Duration of time
Training set 7152 Ten thousand 551 Ten thousand 874 Ten thousand
Test set 361 Ten thousand 27 Ten thousand 43 Ten thousand
TABLE 1
The training set is used as a sample information set to train the initial prediction model to obtain a target prediction model, and the test set is used to test the target prediction model, so that when the exposure (pushing for a host) is about 361 ten thousand times, the click rate of the host is about 27 ten thousand times, and the effective watching time is about 43 ten thousand times.
Compared with the existing prediction model, the model evaluation index AUC (area under the curve, area under curve) of the target prediction model provided by the technical scheme of the application is found to be greatly improved compared with the existing prediction model, wherein the existing prediction model is a model trained by utilizing a plurality of batch sample data, and the batch sample data comprises a plurality of sample information sets with the same sample recommendation information. The larger AUC represents the better training effect of the model and the higher prediction accuracy. See table 2 below:
Model AUC
Existing predictive models 0.8115
Target prediction model 0.8246
TABLE 2
And selecting 10% of users on the line for comparison, and finding that the click rate and the watching time length obtained when the target prediction model provided by the technical scheme of the application is applied are both superior to those of the prior prediction model;
see table 3 below:
TABLE 3 Table 3
Wherein the click rate when the target prediction model is applied is about 1.28% higher than that when the existing prediction model is applied, and the viewing time period when the target prediction model is applied is about 2.47% longer than that when the existing prediction model is applied.
In the embodiment of the application, a plurality of sample information sets can be acquired, each sample information set is input into an initial prediction model, a sample feature set corresponding to each sample information set is generated based on the initial prediction model, the target sample recommendation index of a sample object corresponding to each sample feature set for sample recommendation resources corresponding to the sample object is respectively determined according to each sample feature set, and model parameters of the initial prediction model are corrected according to each target sample recommendation index to obtain a target prediction model. By the method, the target prediction model can have the prediction function of multiple recommended scenes by utilizing the plurality of batch sample data containing different sample scene information for joint training, and learning features under the multiple recommended scenes can be combined during training, so that the prediction efficiency and accuracy of the model are improved.
The foregoing details of the method of embodiments of the present application are provided for the purpose of better implementing the foregoing aspects of embodiments of the present application, and accordingly, the following provides an apparatus of embodiments of the present application.
Fig. 10 is a schematic structural diagram of a data recommendation device provided by the present application, where the data recommendation device may be a computer program (including program code) running in an electronic device, and for example, the data recommendation device may be an application program (such as a program capable of performing data recommendation) in the electronic device. It should be noted that, the data recommending apparatus shown in fig. 10 is configured to perform some or all of the steps in the methods of the embodiments shown in fig. 2, fig. 4, and fig. 8. The data recommendation device 1000 may include an acquisition module 1001, a processing module 1002, and a determination module 1003. Wherein:
An acquisition module 1001, configured to acquire a recommendation information set, where the recommendation information set includes object attribute information of a target object, resource attribute information of a resource to be recommended, and recommendation scene information of the resource to be recommended;
The processing module 1002 is configured to generate a recommendation feature set according to the recommendation information set, where the recommendation feature set includes object attribute features of object attribute information, resource attribute features of resource attribute information, and recommendation scene features of recommendation scene information;
a determining module 1003, configured to determine a first recommendation index of the target object for the resource to be recommended according to the recommendation feature set, where the first recommendation index characterizes a probability that the target object responds to a recommendation behavior for the resource to be recommended;
the determining module 1003 is configured to determine a second recommendation index of the target object for the resource to be recommended under the target scenario information according to the object attribute feature and the resource attribute feature, where the target scenario information is at least one scenario information in a plurality of scenario information associated with the resource to be recommended, and the second recommendation index characterizes a probability that the target object responds to the recommendation behavior for the resource to be recommended;
A determining module 1003, configured to determine a target recommendation index of the target object for the resource to be recommended according to the first recommendation index and the second recommendation index, where the target recommendation index characterizes a target probability of the target object for responding to the recommendation behavior of the resource to be recommended;
The processing module 1002 is configured to push, according to the target recommendation index, the resource to be recommended to the object terminal of the target object.
In one possible implementation manner, the determining module 1003 is specifically configured to, when configured to determine, according to the recommendation feature set, a first recommendation index of the target object for the resource to be recommended:
feature fusion is carried out on the recommended feature set, and fusion attribute features are obtained;
determining a recommendation influence value according to the recommendation characteristic set, wherein the recommendation influence value characterizes the influence degree of the recommendation information set on the response recommendation behavior of the target object;
and determining a first recommendation index according to the fusion attribute characteristics and the recommendation influence values.
In one possible implementation manner, the plurality of scene information includes recommended scene information of the resource to be recommended, and the target scene information is recommended scene information;
the determining module 1003, when configured to determine a second recommendation index of the target object for the resource to be recommended under the target scenario information according to the object attribute feature and the resource attribute feature, is specifically configured to:
acquiring a plurality of scene prediction networks, wherein one scene prediction network corresponds to one scene information;
And predicting object attribute features and resource attribute features according to scene prediction networks corresponding to the recommended scene information in the scene prediction networks to obtain a second recommendation index.
In one possible implementation manner, the plurality of scene information includes recommended scene information of the resource to be recommended, and the target scene information includes the recommended scene information and associated scene information of the recommended scene information in the plurality of scene information;
the determining module 1003, when configured to determine a second recommendation index of the target object for the resource to be recommended under the target scenario information according to the object attribute feature and the resource attribute feature, is specifically configured to:
acquiring a plurality of scene prediction networks, wherein one scene prediction network corresponds to one scene information;
Predicting object attribute characteristics and resource attribute characteristics according to scene prediction networks corresponding to recommended scene information and scene prediction networks corresponding to associated scene information in the scene prediction networks respectively to obtain a first initial recommendation index of a target object for the resource to be recommended under the recommended scene information and a second initial recommendation index of the target object for the resource to be recommended under the associated scene information;
and carrying out weighted summation on the first initial recommendation index and the second initial recommendation index to obtain a second recommendation index.
In one possible implementation, the target recommendation index is generated by a target prediction model, and the processing module 1002 is further configured to:
The method comprises the steps of acquiring a plurality of sample information sets, wherein one sample information set comprises sample object attribute information of a sample object, sample resource attribute information of a sample recommended resource and sample scene information of the sample recommended resource;
inputting each sample information set into an initial prediction model, and generating a sample feature set corresponding to each sample information set based on the initial prediction model, wherein one sample feature set comprises sample object attribute features of sample object attribute information, sample resource attribute features of sample resource attribute information and sample recommended scene features of sample scene information of the sample resource attribute information;
Respectively determining a target sample recommendation index of a sample object corresponding to each sample feature set for a sample recommendation resource corresponding to the sample object according to each sample feature set;
and correcting model parameters of the initial prediction model according to the recommendation index of each target sample to obtain a target prediction model.
In one possible implementation manner, the initial prediction model includes an initial public prediction network and a plurality of initial scene prediction networks, wherein one initial scene prediction network corresponds to one sample of scene information;
the processing module 1002 is specifically configured to, when determining, according to each sample feature set, a target sample recommendation index of a sample object corresponding to each sample feature set for a sample recommendation resource corresponding to the sample object, where the target sample recommendation index is:
In an initial public prediction network, generating a first sample recommendation index of a target sample object for a target sample recommendation resource according to a target feature set, wherein the target feature set is any one sample feature set in each sample feature set, the target sample object is a sample object corresponding to the target feature set, and the target sample recommendation resource is a sample recommendation resource corresponding to the target sample object;
Determining a second sample recommendation index of a target sample object for a target sample recommendation resource according to sample object attribute characteristics in a target characteristic set and sample resource attribute characteristics in the target characteristic set in a target scene prediction network, wherein the target scene prediction network is an initial scene prediction network corresponding to target sample scene information in a plurality of initial scene prediction networks, and the target sample scene information is sample scene information corresponding to the target sample recommendation resource;
And determining a target sample recommendation index of the target sample object for the target sample recommendation resource according to the first sample recommendation index and the second sample recommendation index.
In one possible implementation, the initial prediction model includes an initial public prediction network and a plurality of initial scene prediction networks, and the processing module 1002 is specifically configured to, when modifying model parameters of the initial prediction model according to a recommendation index of each target sample to obtain the target prediction model:
correcting network parameters of an initial public prediction network according to the recommendation index of each target sample to obtain a public prediction network;
Correcting network parameters of a plurality of initial scene prediction networks according to each target sample recommendation index and sample scene information of each sample recommendation resource to obtain a plurality of scene prediction networks;
a target prediction model is determined from the common prediction network and the plurality of scene prediction networks.
In one possible implementation, one initial scene prediction network corresponds to one sample scene information, and the processing module 1002 is specifically configured to, when modifying network parameters of a plurality of initial scene prediction networks according to each target sample recommendation index and sample scene information of each sample recommendation resource to obtain a plurality of scene prediction networks:
Dividing the recommendation index of each target sample into a plurality of initial scene prediction networks according to the sample scene information of each sample recommendation resource;
and correcting the network parameters of the divided initial scene prediction network according to the recommendation index of each target sample to obtain a plurality of scene prediction networks.
According to an embodiment of the present application, each unit in the data recommendation device shown in fig. 10 may be separately or completely combined into one or several other units, or some unit(s) thereof may be further split into a plurality of units with smaller functions, which may achieve the same operation without affecting the implementation of the technical effects of the embodiment of the present application. The above units are divided based on logic functions, and in practical applications, the functions of one unit may be implemented by a plurality of units, or the functions of a plurality of units may be implemented by one unit.
In other embodiments of the present application, the data recommendation device may also include other units, and in practical applications, these functions may also be implemented with assistance from other units, and may be implemented by cooperation of multiple units. According to another embodiment of the present application, a data recommendation apparatus as shown in fig. 10 may be constructed by running a computer program (including program code) capable of executing the steps involved in the respective methods as shown in fig. 2, 4 and 8 on a general-purpose computing device such as a computer including a processing element such as a Central Processing Unit (CPU), a random access storage medium (RAM), a read only storage medium (ROM), and the like, and a storage element, and the data recommendation method of the embodiment of the present application is implemented. The computer program may be recorded on, for example, a computer-readable recording medium, and loaded into and run in the above-described electronic device through the computer-readable recording medium.
In the embodiment of the application, an acquisition module acquires a recommendation information set, a processing module generates a recommendation feature set according to the recommendation information set, a determining module determines a first recommendation index of a target object for a resource to be recommended according to the recommendation feature set, a determining module determines a second recommendation index of the target object for the resource to be recommended under target scene information according to object attribute features and resource attribute features, a determining module determines a target recommendation index of the target object for the resource to be recommended according to the first recommendation index and the second recommendation index, and the processing module pushes the resource to be recommended to an object terminal of the target object according to the target recommendation index. Through the device, the obtained first recommendation index reflects interaction of multiple recommendation features in the recommendation feature set, the obtained second recommendation index is obtained by utilizing the learning features under the target scene information, the learning features under the target scene information can be obtained through the combination of the learning features of all the scene information, the obtained target recommendation index can be combined with multiple recommendation features of the user, and can be combined with the learning features under multiple recommendation scenes, so that recommendation prediction efficiency and accuracy of resources to be predicted can be improved, and the follow-up recommendation effect based on the target recommendation index is facilitated.
The functional modules in the embodiments of the present application may be integrated into one processing module, or each module may exist alone physically, or two or more modules may be integrated into one module. The integrated modules may be implemented in hardware or in software functional modules, which is not limited by the present application.
Referring to fig. 11, fig. 11 is a schematic structural diagram of an electronic device according to an embodiment of the present application. As shown in fig. 11, the electronic device 1100 includes a processor 1101, a communication interface 1102, and a computer-readable storage medium 1103. Wherein the processor 1101, the communication interface 1102, and the computer-readable storage medium 1103 may be connected by a bus or other means. Wherein the communication interface 1102 is used to receive and transmit data. The computer readable storage medium 1103 may be stored in a memory of an electronic device, the computer readable storage medium 1103 being for storing a computer program comprising program instructions, the processor 1101 being for executing the program instructions stored by the computer readable storage medium 1103. The processor 1101 (or CPU (Central Processing Unit, central processing unit)) is a computing core as well as a control core of the electronic device, which is adapted to implement one or more instructions, in particular to load and execute one or more instructions to implement a corresponding method flow or a corresponding function.
The embodiment of the application also provides a computer readable storage medium (Memory), which is a Memory device in the electronic device and is used for storing programs and data. It is understood that the computer readable storage medium herein may include both built-in storage media in an electronic device and extended storage media supported by the electronic device. The computer readable storage medium provides a memory space that stores a processing system of the electronic device. Also stored in this memory space are one or more instructions, which may be one or more computer programs (including program code), adapted to be loaded and executed by the processor 1101. The computer readable storage medium may be a high-speed RAM memory, or may be a non-volatile memory, such as at least one magnetic disk memory, or may alternatively be at least one computer readable storage medium located remotely from the processor.
In one embodiment, the one or more instructions are stored in the computer readable storage medium, the one or more instructions stored in the computer readable storage medium are loaded and executed by the processor 1101 to implement the corresponding steps in the above-mentioned document processing method embodiment, and in a specific implementation, the one or more instructions in the computer readable storage medium are loaded and executed by the processor 1101 as follows:
the method comprises the steps of acquiring a recommendation information set, wherein the recommendation information set comprises object attribute information of a target object, resource attribute information of a resource to be recommended and recommendation scene information of the resource to be recommended;
generating a recommendation feature set according to the recommendation information set, wherein the recommendation feature set comprises object attribute features of object attribute information, resource attribute features of resource attribute information and recommendation scene features of recommendation scene information;
determining a first recommendation index of the target object for the resource to be recommended according to the recommendation feature set, wherein the first recommendation index represents the probability that the target object responds to the recommendation behavior of the resource to be recommended;
determining a second recommendation index of the target object for the resource to be recommended under the target scene information according to the object attribute characteristics and the resource attribute characteristics, wherein the target scene information is at least one scene information in a plurality of scene information associated with the resource to be recommended, and the second recommendation index characterizes the probability that the target object responds to the recommendation behavior of the resource to be recommended;
determining a target recommendation index of a target object for the resource to be recommended according to the first recommendation index and the second recommendation index, pushing the resource to be recommended to an object terminal of the target object according to the target recommendation index, and representing target probability of response recommendation behaviors of the target object to the resource to be recommended.
In one possible implementation, the processor 1101 is specifically configured to, when configured to determine, according to the recommendation feature set, a first recommendation index of the target object for the resource to be recommended:
feature fusion is carried out on the recommended feature set, and fusion attribute features are obtained;
determining a recommendation influence value according to the recommendation characteristic set, wherein the recommendation influence value characterizes the influence degree of the recommendation information set on the response recommendation behavior of the target object;
and determining a first recommendation index according to the fusion attribute characteristics and the recommendation influence values.
In one possible implementation manner, the plurality of scene information includes recommended scene information of the resource to be recommended, and the target scene information is recommended scene information;
The processor 1101 is specifically configured to, when configured to determine a second recommendation index for a resource to be recommended for a target object under the target scenario information according to the object attribute feature and the resource attribute feature:
acquiring a plurality of scene prediction networks, wherein one scene prediction network corresponds to one scene information;
And predicting object attribute features and resource attribute features according to scene prediction networks corresponding to the recommended scene information in the scene prediction networks to obtain a second recommendation index.
In one possible implementation manner, the plurality of scene information includes recommended scene information of the resource to be recommended, and the target scene information includes the recommended scene information and associated scene information of the recommended scene information in the plurality of scene information;
The processor 1101 is specifically configured to, when configured to determine a second recommendation index for a resource to be recommended for a target object under the target scenario information according to the object attribute feature and the resource attribute feature:
acquiring a plurality of scene prediction networks, wherein one scene prediction network corresponds to one scene information;
Predicting object attribute characteristics and resource attribute characteristics according to scene prediction networks corresponding to recommended scene information and scene prediction networks corresponding to associated scene information in the scene prediction networks respectively to obtain a first initial recommendation index of a target object for the resource to be recommended under the recommended scene information and a second initial recommendation index of the target object for the resource to be recommended under the associated scene information;
and carrying out weighted summation on the first initial recommendation index and the second initial recommendation index to obtain a second recommendation index.
In one possible implementation, the target recommendation index is generated by a target prediction model, and the processor 1101 is further configured to:
The method comprises the steps of acquiring a plurality of sample information sets, wherein one sample information set comprises sample object attribute information of a sample object, sample resource attribute information of a sample recommended resource and sample scene information of the sample recommended resource;
inputting each sample information set into an initial prediction model, and generating a sample feature set corresponding to each sample information set based on the initial prediction model, wherein one sample feature set comprises sample object attribute features of sample object attribute information, sample resource attribute features of sample resource attribute information and sample recommended scene features of sample scene information of the sample resource attribute information;
Respectively determining a target sample recommendation index of a sample object corresponding to each sample feature set for a sample recommendation resource corresponding to the sample object according to each sample feature set;
and correcting model parameters of the initial prediction model according to the recommendation index of each target sample to obtain a target prediction model.
In one possible implementation manner, the initial prediction model includes an initial public prediction network and a plurality of initial scene prediction networks, wherein one initial scene prediction network corresponds to one sample of scene information;
The processor 1101 is specifically configured to, when configured to determine, according to each sample feature set, a target sample recommendation index of a sample object corresponding to each sample feature set for a sample recommendation resource corresponding to the sample object, respectively:
In an initial public prediction network, generating a first sample recommendation index of a target sample object for a target sample recommendation resource according to a target feature set, wherein the target feature set is any one sample feature set in each sample feature set, the target sample object is a sample object corresponding to the target feature set, and the target sample recommendation resource is a sample recommendation resource corresponding to the target sample object;
Determining a second sample recommendation index of a target sample object for a target sample recommendation resource according to sample object attribute characteristics in a target characteristic set and sample resource attribute characteristics in the target characteristic set in a target scene prediction network, wherein the target scene prediction network is an initial scene prediction network corresponding to target sample scene information in a plurality of initial scene prediction networks, and the target sample scene information is sample scene information corresponding to the target sample recommendation resource;
And determining a target sample recommendation index of the target sample object for the target sample recommendation resource according to the first sample recommendation index and the second sample recommendation index.
In one possible implementation, the initial prediction model includes an initial public prediction network and a plurality of initial scene prediction networks, and the processor 1101 is configured to, when modifying model parameters of the initial prediction model according to a recommendation index of each target sample, obtain the target prediction model, specifically:
correcting network parameters of an initial public prediction network according to the recommendation index of each target sample to obtain a public prediction network;
Correcting network parameters of a plurality of initial scene prediction networks according to each target sample recommendation index and sample scene information of each sample recommendation resource to obtain a plurality of scene prediction networks;
a target prediction model is determined from the common prediction network and the plurality of scene prediction networks.
In one possible implementation, one initial scene prediction network corresponds to one sample scene information, and the processor 1101 is configured to, when modifying network parameters of a plurality of initial scene prediction networks according to each target sample recommendation index and the sample scene information of each sample recommendation resource to obtain a plurality of scene prediction networks, specifically:
Dividing the recommendation index of each target sample into a plurality of initial scene prediction networks according to the sample scene information of each sample recommendation resource;
and correcting the network parameters of the divided initial scene prediction network according to the recommendation index of each target sample to obtain a plurality of scene prediction networks.
In the embodiment of the application, the processor can acquire a recommendation information set, generate a recommendation feature set according to the recommendation information set, determine a first recommendation index of a target object for a resource to be recommended according to the recommendation feature set, the first recommendation index represents the probability of response recommendation behavior of the target object for the resource to be recommended, determine a second recommendation index of the target object for the resource to be recommended under the condition of target scene information according to the object attribute features and the resource attribute features, determine a target recommendation index of the target object for the resource to be recommended according to the first recommendation index and the second recommendation index, and push the resource to be recommended to an object terminal of the target object according to the target recommendation index. Through the scheme, the obtained first recommendation index reflects interaction of multiple recommendation features in the recommendation feature set, the obtained second recommendation index is obtained by utilizing learning features under target scene information, the learning features under the target scene information can be obtained through combination of the learning features of all scene information, the obtained target recommendation index can be combined with multiple recommendation features of the user in a deeper level, and can be combined with the learning features under multiple recommendation scenes, so that recommendation prediction efficiency and accuracy of resources to be predicted can be improved, and the follow-up recommendation effect based on the target recommendation index is facilitated to be improved.
Embodiments of the present application also provide a computer program product comprising program instructions which, when executed by a processor, implement some or all of the steps of the above-described method.
Those of ordinary skill in the art will appreciate that the various illustrative elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware, or combinations of computer software and electronic hardware. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the solution. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present application.
In the above embodiments, it may be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented in software, may be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the processes or functions described in accordance with embodiments of the present invention are produced in whole or in part. The computer may be a general purpose computer, a special purpose computer, a network of computers, or other programmable devices. The computer instructions may be stored in or transmitted across a computer-readable storage medium. The computer instructions may be transmitted from one website, computer, server, or data center to another website, computer, server, or data center by a wired (e.g., coaxial cable, fiber optic, digital object line (DSL)), or wireless (e.g., infrared, wireless, microwave, etc.). Computer readable storage media can be any available media that can be accessed by a computer or data storage devices, such as servers, data centers, etc., that contain an integration of one or more available media. The usable medium may be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid state disk (Solid STATE DISK, SSD)), or the like.
The foregoing is merely illustrative of the present application, and the present application is not limited thereto, and any person skilled in the art will readily recognize that variations or substitutions are within the scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.

Claims (11)

1. A data recommendation method, the method comprising:
The method comprises the steps of acquiring a recommendation information set, wherein the recommendation information set comprises object attribute information of a target object, resource attribute information of a resource to be recommended and recommendation scene information of the resource to be recommended;
Generating a recommendation feature set according to the recommendation information set, wherein the recommendation feature set comprises object attribute features of the object attribute information, resource attribute features of the resource attribute information and recommendation scene features of the recommendation scene information;
Determining a first recommendation index of the target object for the resource to be recommended according to the recommendation feature set, wherein the first recommendation index represents the probability of the response recommendation behavior of the target object for the resource to be recommended;
Determining a second recommendation index of the target object for the resource to be recommended under target scene information according to the object attribute characteristics and the resource attribute characteristics, wherein the target scene information is at least one scene information in a plurality of scene information associated with the resource to be recommended, and the second recommendation index characterizes the probability of the target object to respond to the recommended behavior of the resource to be recommended;
The target recommendation index of the target object for the resource to be recommended is determined according to the first recommendation index and the second recommendation index, the resource to be recommended is pushed to an object terminal of the target object according to the target recommendation index, the target recommendation index represents target probability of response recommendation behavior of the target object for the resource to be recommended, the target recommendation index is generated by a target prediction model, the target prediction model is obtained by training an initial prediction model by utilizing a target sample recommendation index of a sample object corresponding to a sample information set for the sample recommendation resource corresponding to the sample object, the target sample recommendation index corresponding to the sample information set is obtained based on a sample feature set corresponding to the sample information set, the sample information set comprises sample object attribute information of a sample object, sample resource attribute information of the sample recommendation resource and sample scene information of the sample recommendation resource, and the sample feature set comprises sample object attribute characteristics of sample object attribute information, sample resource attribute characteristics of sample resource attribute information of the sample resource attribute information and sample scene characteristics of the sample scene information of the sample resource attribute information.
2. The method of claim 1, wherein the determining a first recommendation index for the target object for the resource to be recommended based on the set of recommendation characteristics comprises:
feature fusion is carried out on the recommended feature set, and fusion attribute features are obtained;
Determining a recommendation influence value according to the recommendation characteristic set, wherein the recommendation influence value characterizes the influence degree of the recommendation information set on the target object response recommendation behavior;
And determining the first recommendation index according to the fusion attribute characteristics and the recommendation influence value.
3. The method of claim 1, wherein the plurality of scene information includes recommended scene information of a resource to be recommended, the target scene information being the recommended scene information;
determining a second recommendation index of the target object for the resource to be recommended under the target scene information according to the object attribute characteristics and the resource attribute characteristics, wherein the second recommendation index comprises the following components:
acquiring a plurality of scene prediction networks, wherein one scene prediction network corresponds to one scene information;
and predicting the object attribute characteristics and the resource attribute characteristics according to a scene prediction network corresponding to the recommended scene information in the scene prediction networks to obtain the second recommendation index.
4. The method of claim 1, wherein the plurality of scene information includes recommended scene information of a resource to be recommended, and the target scene information includes the recommended scene information and associated scene information of the recommended scene information of the plurality of scene information;
determining a second recommendation index of the target object for the resource to be recommended under the target scene information according to the object attribute characteristics and the resource attribute characteristics, wherein the second recommendation index comprises the following components:
acquiring a plurality of scene prediction networks, wherein one scene prediction network corresponds to one scene information;
Predicting the object attribute characteristics and the resource attribute characteristics according to a scene prediction network corresponding to the recommended scene information and a scene prediction network corresponding to the associated scene information in the plurality of scene prediction networks respectively to obtain a first initial recommendation index of the target object for the resource to be recommended under the recommended scene information and a second initial recommendation index of the target object for the resource to be recommended under the associated scene information;
and carrying out weighted summation on the first initial recommendation index and the second initial recommendation index to obtain the second recommendation index.
5. The method according to claim 1, wherein the method further comprises:
Acquiring a plurality of sample information sets;
Inputting each sample information set into an initial prediction model, and generating a sample feature set corresponding to each sample information set based on the initial prediction model;
Respectively determining a target sample recommendation index of a sample object corresponding to each sample feature set for a sample recommendation resource corresponding to the sample object according to each sample feature set;
And correcting model parameters of the initial prediction model according to the recommendation index of each target sample to obtain the target prediction model.
6. The method of claim 5, wherein the initial predictive model includes an initial common predictive network and a plurality of initial scene predictive networks, one initial scene predictive network corresponding to each piece of scene information;
The determining, according to each sample feature set, a target sample recommendation index of a sample object corresponding to each sample feature set for a sample recommendation resource corresponding to the sample object, includes:
In the initial public prediction network, generating a first sample recommendation index of a target sample object for a target sample recommendation resource according to a target feature set, wherein the target feature set is any one of the sample feature sets, the target sample object is a sample object corresponding to the target feature set, and the target sample recommendation resource is a sample recommendation resource corresponding to the target sample object;
Determining a second sample recommendation index of the target sample object for the target sample recommendation resource according to sample object attribute characteristics in the target characteristic set and sample resource attribute characteristics in the target characteristic set in a target scene prediction network, wherein the target scene prediction network is an initial scene prediction network corresponding to target sample scene information in the plurality of initial scene prediction networks, and the target sample scene information is sample scene information corresponding to the target sample recommendation resource;
And determining a target sample recommendation index of the target sample object for the target sample recommendation resource according to the first sample recommendation index and the second sample recommendation index.
7. The method of claim 5, wherein the initial predictive model includes an initial common predictive network and a plurality of initial scene predictive networks, wherein modifying model parameters of the initial predictive model based on each target sample recommendation index to obtain the target predictive model includes:
Correcting network parameters of the initial public prediction network according to the recommendation index of each target sample to obtain a public prediction network;
Correcting network parameters of the plurality of initial scene prediction networks according to the target sample recommendation indexes and sample scene information of the sample recommendation resources to obtain a plurality of scene prediction networks;
the target prediction model is determined from the common prediction network and the plurality of scene prediction networks.
8. The method of claim 7, wherein one initial scene prediction network corresponds to one sample scene information, wherein the correcting network parameters of the plurality of initial scene prediction networks according to the sample scene information of each target sample recommendation index and each sample recommendation resource to obtain a plurality of scene prediction networks comprises:
Dividing the recommendation index of each target sample into a plurality of initial scene prediction networks according to the sample scene information of each sample recommendation resource;
And correcting the network parameters of the divided initial scene prediction networks according to the recommendation index of each target sample to obtain the scene prediction networks.
9. An electronic device comprising a processor and a memory, wherein the memory is configured to store a computer program comprising program instructions, the processor being configured to invoke the program instructions to perform the method of any of claims 1-8.
10. A computer readable storage medium, characterized in that the computer readable storage medium stores a computer program comprising program instructions which, when executed by a processor, cause the processor to perform the method of any of claims 1-8.
11. A computer program product, characterized in that it stores program instructions that, when executed by a processor, cause the processor to perform the method according to any of claims 1-8.
CN202210436960.7A 2022-04-24 2022-04-24 Data recommendation method, electronic device and storage medium Active CN115033777B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210436960.7A CN115033777B (en) 2022-04-24 2022-04-24 Data recommendation method, electronic device and storage medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210436960.7A CN115033777B (en) 2022-04-24 2022-04-24 Data recommendation method, electronic device and storage medium

Publications (2)

Publication Number Publication Date
CN115033777A CN115033777A (en) 2022-09-09
CN115033777B true CN115033777B (en) 2024-12-20

Family

ID=83118692

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210436960.7A Active CN115033777B (en) 2022-04-24 2022-04-24 Data recommendation method, electronic device and storage medium

Country Status (1)

Country Link
CN (1) CN115033777B (en)

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113536138A (en) * 2021-08-16 2021-10-22 北京爱奇艺科技有限公司 Network resource recommendation method and device, electronic equipment and readable storage medium
CN113742590A (en) * 2021-09-07 2021-12-03 北京沃东天骏信息技术有限公司 Recommendation method and device, storage medium and electronic equipment

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2018095366A1 (en) * 2016-11-24 2018-05-31 腾讯科技（深圳）有限公司 Frame-synchronisation-based data processing method for video recommendation determination and information display
CN113886674B (en) * 2020-07-01 2024-09-06 北京达佳互联信息技术有限公司 Resource recommendation method and device, electronic equipment and storage medium

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113536138A (en) * 2021-08-16 2021-10-22 北京爱奇艺科技有限公司 Network resource recommendation method and device, electronic equipment and readable storage medium
CN113742590A (en) * 2021-09-07 2021-12-03 北京沃东天骏信息技术有限公司 Recommendation method and device, storage medium and electronic equipment

Also Published As

Publication number Publication date
CN115033777A (en) 2022-09-09

Similar Documents

Publication Publication Date Title
CN110941740B (en) Video recommendation method and computer-readable storage medium
WO2022022152A1 (en) Video clip positioning method and apparatus, and computer device and storage medium
US11868738B2 (en) Method and apparatus for generating natural language description information
CN114265979B (en) Method for determining fusion parameters, information recommendation method and model training method
CN112149604B (en) Video feature extraction model training method, video recommendation method and device
CN111159563A (en) Method, device and equipment for determining user interest point information and storage medium
CN108475256A (en) Feature insertion is generated from homologous factors
CN115455280A (en) Recommendation list determining method and server
CN112883257A (en) Behavior sequence data processing method and device, electronic equipment and storage medium
CN111814044B (en) Recommendation method, recommendation device, terminal equipment and storage medium
CN117688204A (en) Training method and device for video recommendation model, electronic equipment and storage medium
CN115514995B (en) A method, device and equipment for displaying recommended information in a live broadcast room
CN115129945A (en) Graph structure contrast learning method, equipment and computer storage medium
CN115033777B (en) Data recommendation method, electronic device and storage medium
CN115017362A (en) Data processing method, electronic device and storage medium
CN112100511A (en) Preference degree data obtaining method and device and electronic equipment
CN117216361A (en) Recommendation method, recommendation device, electronic equipment and computer readable storage medium
CN115080856A (en) Recommended method and device, training method and device of recommended model
CN115115916A (en) Feature extraction method, feature extraction device, computer equipment and storage medium
CN116777529B (en) Object recommendation method, device, equipment, storage medium and program product
CN116017068B (en) Live broadcast recommendation method, device, equipment and computer-readable storage medium
KR102650587B1 (en) Method, device and system for providing evaluation and matching solutions for contract brokers considering quantitative and qualitative criteria
CN113742531B (en) Picture recommendation method and device and electronic equipment
CN116662637B (en) Content recommendation method, device, equipment, storage medium and program product
CN117312646A (en) Media diversity recommendation method, device, computer equipment and storage medium

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant