Object recommendation method, object recommendation device, electronic device and storage medium

Abstract

The present application discloses an object recommendation method, an object recommendation device, an electronic device, and a storage medium, belonging to the field of artificial intelligence technology. The method includes: obtaining reference information, the reference information including scene information of multiple scenes, or the reference information including user information of a target user, at least one of object information of multiple objects, and scene information of multiple scenes; determining reference features based on the reference information, the reference features including scene features of each scene, or the reference features including user features of the target user, at least one of object features of each object, and scene features of each scene; determining an object to be recommended from multiple objects based on the reference features; and recommending the object to be recommended to the target user. This method realizes the determination of the object to be recommended based on scene information, or the integration of user information, at least one of object information, and scene information, thereby improving accuracy, and increasing the usage time and frequency of the application.

Description

Technical Field

[0001] The embodiments of the present application relate to the field of artificial intelligence technology, and in particular to an object recommendation method, an object recommendation device, an electronic device, and a storage medium. Background Art

[0002] With the development of artificial intelligence technology, the functions of applications are becoming more and more powerful. Many applications on the market have recommendation functions, such as video applications with video recommendation functions and food applications with food recommendation functions.

[0003] In the related art, there is a class of applications that feature object recommendation. These applications recommend objects previously selected by a user to the user, allowing the user to first select a target object from the recommended objects and then select a target resource from the resources published by the target object. However, since object recommendations are based solely on the user's previously selected objects (i.e., user and object information), accuracy is low, impacting user usage of the application. Summary of the Invention

[0004] The embodiments of the present application provide an object recommendation method, an object recommendation device, an electronic device, and a storage medium, which can be used to solve problems in related technologies. The technical solution includes the following contents.

[0005] In one aspect, an embodiment of the present application provides an object recommendation method, the method comprising:

[0006] Acquiring reference information, where the reference information includes scene information of multiple scenes, or the reference information includes user information of a target user, at least one of object information of multiple objects, and scene information of the multiple scenes, where the scene represents a type of interactive behavior, where the interactive behavior is an action of a user selecting a resource published by an object in an environment;

[0007] Determining reference features based on the reference information, where the reference features include scene features of each scene, or the reference features include at least one of a user feature of the target user and an object feature of each object, and the scene features of each scene;

[0008] Determining an object to be recommended from the multiple objects based on the reference features;

[0009] Recommend the object to be recommended to the target user.

[0010] On the other hand, an embodiment of the present application provides an object recommendation device, the device comprising:

[0011] an acquisition module, configured to acquire reference information, the reference information including scene information of multiple scenes, or the reference information including user information of a target user, at least one of object information of multiple objects, and scene information of the multiple scenes, the scene representing a type of interactive behavior, the interactive behavior being an action of a user selecting a resource published by an object in an environment;

[0012] a determination module, configured to determine reference features based on the reference information, wherein the reference features include scene features of each scene, or the reference features include at least one of a user feature of the target user and an object feature of each object, and the scene features of each scene;

[0013] The determining module is further configured to determine an object to be recommended from the plurality of objects based on the reference features;

[0014] The recommendation module is used to recommend the object to be recommended to the target user.

[0015] On the other hand, an embodiment of the present application provides an electronic device, which includes a processor and a memory, wherein the memory stores at least one program code, and the at least one program code is loaded and executed by the processor so that the electronic device implements any of the object recommendation methods described above.

[0016] On the other hand, a computer-readable storage medium is provided, in which at least one program code is stored. The at least one program code is loaded and executed by a processor to enable a computer to implement any of the above-mentioned object recommendation methods.

[0017] On the other hand, a computer program or a computer program product is also provided, wherein the computer program or the computer program product stores at least one computer instruction, and the at least one computer instruction is loaded and executed by a processor to enable a computer to implement any of the above-mentioned object recommendation methods.

[0018] The technical solutions provided by the embodiments of the present application bring at least the following beneficial effects:

[0019] The technical solution provided in the embodiments of the present application is to determine the object to be recommended from multiple objects based on the scene characteristics of each scene, or based on the user characteristics of the target user, at least one of the object characteristics of each object and the scene characteristics of each scene, thereby realizing the determination of the object to be recommended based on scene information or comprehensive user information, at least one of the object information and scene information, thereby improving accuracy and thus increasing the time and frequency of users using the application. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0021] Figure 1 This is a schematic diagram of an implementation environment of an object recommendation method provided in an embodiment of the present application;

[0022] Figure 2 This is a flowchart of an object recommendation method provided by an embodiment of the present application;

[0023] Figure 3 This is a schematic diagram of a scenario provided by an embodiment of the present application;

[0024] Figure 4 This is another scenario diagram provided by an embodiment of the present application;

[0025] Figure 5 is a schematic diagram of a heterogeneous scene hypergraph provided in an embodiment of the present application;

[0026] Figure 6 This is a schematic diagram of determining an object to be recommended provided by an embodiment of the present application;

[0027] Figure 7 Schematic diagram of the structure of an object recommendation device provided in an embodiment of the present application;

[0028] Figure 8 This is a schematic diagram of the structure of a terminal device provided in an embodiment of the present application;

[0029] Figure 9 This is a structural diagram of a server provided in an embodiment of the present application. DETAILED DESCRIPTION

[0030] In order to make the objectives, technical solutions and advantages of this application clearer, the implementation methods of this application will be further described in detail below with reference to the accompanying drawings.

[0031] Figure 1 This is a schematic diagram of an implementation environment of an object recommendation method provided in an embodiment of the present application, such as Figure 1 The implementation environment shown includes an electronic device 11, and the object recommendation method in the embodiment of the present application can be executed by the electronic device 11. Exemplarily, the electronic device 11 can include at least one of a terminal device and a server.

[0032] The terminal device may be at least one of a smartphone, a desktop computer, a tablet computer, an e-book reader, and a laptop computer. The server may be a single server, a server cluster consisting of multiple servers, or any one of a cloud computing platform and a virtualization center, which are not limited in the embodiments of the present application. The server may be connected to the terminal device via a wired network or a wireless network. The server may have functions such as data processing, data storage, and data transmission and reception, which are not limited in the embodiments of the present application.

[0033] Based on the above implementation environment, the present application embodiment provides an object recommendation method. Figure 2 As an example, the flowchart of an object recommendation method provided in the embodiment of the present application is shown. The method can be performed by Figure 1 The electronic device 11 in the embodiment is executed. Figure 2 As shown, the method includes steps 201 to 204.

[0034] Step 201, obtain reference information, the reference information includes scene information of multiple scenes, or the reference information includes user information of the target user, at least one item of object information of multiple objects and scene information of multiple scenes, the scene characterizes the type of interactive behavior, and the interactive behavior is the behavior of the user selecting resources published by the object in the environment.

[0035] The embodiment of the present application does not limit the number of target users. The target user can be any one user or multiple users. The user information includes at least one of the following attribute information: gender, age, birthday, height, etc.

[0036] An object may be referred to as a point of interest (POI). In the embodiment of the present application, the object may be a store, in which case the object information includes at least one of the following attributes: store area, store location, and industry. In the embodiment of the present application, the object may also be a media uploader (Uploader), in which case the object information includes at least one of the following attributes: gender, account name, and profile.

[0037] A scenario represents the type of interactive behavior. Interaction occurs when a user selects resources published by an object within an environment. The environment refers to the user's surroundings, which can be represented by environmental information. For example, environmental information includes at least one of the following attributes: time, location, and weather. If the object is a store, the resources published by the object can be items. If the object is a media uploader, the resources published by the object can be media information.

[0038] See Figure 3 , Figure 3 This is a scenario diagram provided in an embodiment of the present application. Figure 3 There are two interactive behaviors, which are respectively recorded as interactive behavior 1 and interactive behavior 2. Interactive behavior 1 is: Zhang San (i.e., user) selects pizza and juice (i.e., resources) posted by a pizza shop (i.e., object) in environments such as dinner, office building, and weekday. Interactive behavior 2 is: Li Si (i.e., user) selects pasta (i.e., resource) posted by a pizza shop (i.e., object) in environments such as lunch and office building. "Scenario 1: weekday, fast food" can be used to characterize the type of interactive behavior 1 and the type of interactive behavior 2. It can be understood that in addition to corresponding to interactive behavior 1 and interactive behavior 2, scenario 1 can also correspond to other interactive behaviors, and the embodiment of the present application does not limit the number of interactive behaviors corresponding to scenario 1.

[0039] Next, see Figure 4 , Figure 4 This is another scenario diagram provided in an embodiment of the present application. Figure 4 There are two interactive behaviors, which are respectively recorded as interactive behavior 3 and interactive behavior 4. Interactive behavior 3 is: Zhang San (i.e., the user) selects coffee and juice (i.e., resources) posted by a coffee shop (i.e., the object) in the afternoon, office building, and other environments. Interactive behavior 4 is: Wang Wu (i.e., the user) selects milk tea (i.e., resources) posted by a milk tea shop (i.e., the object) in the afternoon, school, and other environments. "Scene 2: Afternoon, Wake Up" can be used to characterize the types of interactive behavior 3 and interactive behavior 4. It can be understood that in addition to corresponding to interactive behaviors 3 and interactive behaviors 4, scene 2 can also correspond to other interactive behaviors, and the embodiment of the present application does not limit the number of interactive behaviors corresponding to scene 2.

[0040] It should be noted that the scenarios in the embodiments of this application correspond to at least one interactive behavior. A scenario is determined based on an interactive behavior. It is obtained by abstracting and summarizing the user, environment, object, and resources involved in the interactive behavior, and is used to characterize the type of the interactive behavior. Therefore, scenario information includes at least one of the following attribute information: user type, environment type, object type, resource type, etc.

[0041] Alternatively, given a four-tuple <U, P, I, C>, where U represents the user information of multiple users, P represents the object information of multiple objects, I represents the resource information of multiple resources, and C represents the environmental information of multiple environments, an interaction behavior can be defined as: τ = <u, p, i, c>. Here, τ represents the interaction behavior, u∈U represents the user information of a user, p∈P represents the object information of the user's selected object, i∈I represents the resource information of the resource published by the user's selected object, and c∈C represents the environmental information of the environment.

[0042] For an interactive behavior, a scene can be determined based on the interactive behavior. The scene can be expressed as: s = ψ(τ). Where s represents the scene, ψ represents the scene function, and the scene function is used to determine a scene based on an interactive behavior. τ represents the interactive behavior.

[0043] In an embodiment of the present application, scene information of multiple scenes can be obtained, and user information of a target user, at least one item of object information of multiple objects, and scene information of multiple scenes can also be obtained.

[0044] Optionally, user information of the target user, object information of multiple objects, and scene information of multiple scenes can be obtained. A medium can be used to store the user information of the target user, the object information of multiple objects, and the scene information of multiple scenes. By obtaining the medium, the user information of the target user, the object information of multiple objects, and the scene information of multiple scenes on the medium can be obtained. The embodiments of the present application do not limit the medium. Exemplarily, the medium can be a heterogeneous scene hypergraph. For ease of description, each optional embodiment below is explained using the example of a medium being a heterogeneous scene hypergraph.

[0045] The heterogeneous scene hypergraph includes user nodes, object nodes, and scene edges. There are multiple user nodes, each representing user information for a user. The multiple user nodes include a target user node, which represents user information for the target user. There are also multiple object nodes, each representing object information for an object. There are also multiple scene edges, each representing scene information for a scene. Optionally, a scene edge can be a closed graph. Since the scene edge represents the type of an interactive behavior, the closed graph includes the user node corresponding to the user and the object node corresponding to the object in the interactive behavior.

[0046] In an embodiment of the present application, a heterogeneous scene hypergraph can be constructed based on user information of multiple users, object information of multiple objects, and scene information of multiple scenes. By obtaining the heterogeneous scene hypergraph, it is possible to obtain user information of the target user, object information of multiple objects, and scene information of multiple scenes on the heterogeneous scene hypergraph.

[0047] Step 202 : Determine reference features based on reference information, where the reference features include scene features of each scene, or the reference features include user features of a target user, at least one of object features of each object, and scene features of each scene.

[0048] In the embodiment of the present application, the user characteristics of the target user are determined based on the user information of the target user, the object characteristics of each object are determined based on the object information of each object, and the scene characteristics of each scene are determined based on the scene information of each scene.

[0049] Optionally, the heterogeneous scene hypergraph includes user information of the target user, object information of multiple objects, and scene information of multiple scenes. The user characteristics of the target user, the object characteristics of each object, and the scene characteristics of each scene can be determined based on the heterogeneous scene hypergraph.

[0050] The following will introduce the methods for determining the user characteristics of the target user (see implementation methods A1-A3 for details), the method for determining the object characteristics of each object (see implementation methods B1-B4 for details), and the method for determining the scene characteristics of each scene (see implementation methods C1-C2 for details).

[0051] Implementation method A1, the reference information includes user information of the target user, and the reference feature is determined based on the reference information, including: determining the first feature of the target user based on the user information of the target user; obtaining at least one first association information, and determining the second feature of the target user based on the at least one first association information, any first association information includes object information of any object selected by the target user; determining the user feature of the target user based on the first feature of the target user and the second feature of the target user.

[0052] In an embodiment of the present application, a heterogeneous scenario hypergraph includes user information of multiple users. User information of a target user can be first determined from the user information of the multiple users, where the target user is at least one user from the multiple users. Then, a first feature of the target user is determined based on the user information of the target user. The user information of the target user includes at least one attribute information. The first feature of the target user can be determined based on the user information of the target user according to Formula (1) shown below.

[0053] u=h(e1,e2,…,e F ) Formula (1)

[0054] Among them, u represents the first feature of the target user, h represents the aggregation function, e1, e2, ..., e F They represent the features corresponding to the attribute information, and F represents the number of attribute information. Optionally, the aggregation function can be an average function, a layer normalization function, or a composite function of a layer normalization function and an average function, that is, LayerNorm represents the layer normalization function, represents a composite function, and AVG represents an average function.

[0055] Optionally, the feature corresponding to the attribute information is determined according to formula (2) shown below.

[0056] e i =1 / (q·x i ·M i), i=1, 2, …, F Formula (2)

[0057] Among them, e i Represents the feature corresponding to the i-th attribute information, i takes any value from 1 to F, F represents the number of attribute information, q represents x i The number of non-zero elements in , One-Hot Encoding or Multi-Hot Encoding representing the i-th attribute information, R represents a real number, 1×d i Represents the dimension of one-hot encoding or multi-hot encoding, The embedding matrix representing the i-th attribute information, d i ×d e Characterizes the dimensionality of the embedding matrix.

[0058] The heterogeneous scene hypergraph of the present embodiment also includes a selection edge, with a user node and an object node at either end. That is, the heterogeneous scene hypergraph includes a user node, a selection edge, and an object node. The user node, the selection edge, and the object node represent any user selecting any object. A target user node, the selection edge, and the object node can be determined from the heterogeneous scene hypergraph. The target user node is the user node corresponding to the target user.

[0059] The number of target user nodes, selected edges, and object nodes is at least one. For any "target user node, selected edge, and object node," the object information represented by the object node is used as first association information. In this manner, each piece of first association information can be determined.

[0060] In the embodiment of the present application, the first feature of each object selected by the target user can be determined based on each piece of first association information, and the second feature of the target user can be determined based on the first feature of each object selected by the target user.

[0061] The object information includes at least one attribute information, and the first feature of the object can be expressed as p=h(e1, e2, ..., e F ), where p represents the first feature of the object, h represents the aggregation function, e1, e2, ..., e F They respectively represent the features corresponding to the attribute information, and F represents the amount of attribute information. The features corresponding to the attribute information can be determined according to the formula (2) mentioned above.

[0062] In the embodiment of the present application, the second feature of the target user is determined based on the first features of each object selected by the target user according to the formula (3) shown below.

[0063] u p=Transformer(p1,p2,…,p n ) Formula (3)

[0064] Among them, u p The second feature representing the target user, the model parameters of the Transformer representation conversion model (a model of self-attention mechanism), p1, p2, ..., p n Characterizes the first feature of each object selected by the target user.

[0065] After determining the first feature and the second feature of the target user, the first feature and the second feature of the target user can be fused to obtain the user feature of the target user. The embodiment of the present application does not limit the fusion method. For example, the fusion method can be addition, splicing, etc.

[0066] Implementation method A2, the reference information includes user information of the target user, and the reference feature is determined based on the reference information, including: determining the first feature of the target user based on the user information of the target user; obtaining at least one second associated information, and determining the third feature of the target user based on the at least one second associated information, any second associated information including the scene information of any scene and the object information of any object when the target user selects any object in any scene; determining the user feature of the target user based on the first feature of the target user and the third feature of the target user.

[0067] The method of determining the first feature of the target user based on the user information of the target user has been introduced in Implementation A1 and will not be repeated here.

[0068] In the embodiment of the present application, since the heterogeneous scene hypergraph also includes scene edges, which represent the scene information of a scene and can be a closed graph containing user nodes and object nodes, the heterogeneous scene hypergraph includes user nodes, scene edges, and object nodes. The user node, scene edge, and object node represent any user selecting any object in any scene. The target user node, scene edge, and object node can be determined from the heterogeneous scene hypergraph.

[0069] The number of target user nodes, scene edges, and object nodes is at least one. For any target user node, scene edge, and object node, a second association information is determined using the scene information represented by the scene edge and the object information represented by the object node. In this manner, each piece of second association information can be determined.

[0070] In an embodiment of the present application, the first characteristics of each scene in which the target user is located and the first characteristics of each object selected by the target user in each scene can be determined based on each second association information, and the third characteristics of the target user can be determined based on the first characteristics of each scene in which the target user is located and the first characteristics of each object selected by the target user in each scene.

[0071] The scene information includes at least one attribute information, and the first feature of the scene can be expressed as s=h(e1, e2, ..., e F ), where s represents the first feature of the scene, h represents the aggregation function, e1, e2, ..., e F They respectively represent the features corresponding to the attribute information, and F represents the amount of attribute information. The features corresponding to the attribute information can be determined according to the formula (2) mentioned above.

[0072] In the embodiment of the present application, the third feature of the target user is determined based on the first feature of each scene in which the target user is located and the first feature of each object selected by the target user in each scene according to the formula (4) shown below.

[0073] u sp =AVG({p j ⊙s j |j=1,2,…,n}) Formula (4)

[0074] Among them, u sp Characterize the third feature of the target user, AVG characterizes the average function, p j The first feature that characterizes the object selected by the target user in the jth scene, s j The first feature that represents the target user in the jth scene, n is the number of scenes where the target user is located, and ⊙ represents the operator symbol of the Hadamard product. The Hadamard product is an operation that multiplies the corresponding elements of two vectors.

[0075] After determining the first and third characteristics of the target user, the first and third characteristics of the target user can be fused to obtain the user characteristics of the target user. This embodiment of the application does not limit the fusion method. For example, the fusion method can be addition, splicing, etc.

[0076] Implementation method A3, the reference information includes user information of the target user, and the reference feature is determined based on the reference information, including: determining the first feature of the target user based on the user information of the target user; obtaining at least one third related information, and determining the fourth feature of the target user based on the at least one third related information, any third related information including the scene information of any scene and the resource information of any resource when the target user selects any resource in any scene; determining the user feature of the target user based on the first feature of the target user and the fourth feature of the target user.

[0077] The method of determining the first feature of the target user based on the user information of the target user has been introduced in Implementation A1 and will not be repeated here.

[0078] The heterogeneous scenario hypergraph in the embodiments of this application also includes resource nodes. There are multiple resource nodes, each representing resource information for a resource, which includes at least one attribute. In the embodiments of this application, resources are published by an object. For a store, the resources can include clothing, food, or electrical appliances. For a media uploader, the resources can include videos, images, and articles.

[0079] Because scenarios represent the type of user behavior in selecting resources published by objects in an environment, scenario edges can be closed graphs consisting of user nodes, object nodes, and resource nodes. Therefore, the heterogeneous scenario hypergraph includes user nodes, scenario edges, and resource nodes. User nodes, scenario edges, and resource nodes represent any user selecting any resource in any scenario. The target user node, scenario edge, and resource node can be determined from the heterogeneous scenario hypergraph.

[0080] The number of target user nodes, scenario edges, and resource nodes is at least one. For any target user node, scenario edge, and resource node, a third association information is determined using the scenario information represented by the scenario edge and the resource information represented by the resource node. In this manner, each third association information can be determined.

[0081] In an embodiment of the present application, the first characteristics of each scene in which the target user is located and the first characteristics of each resource selected by the target user in each scene can be determined based on each third association information, and the fourth characteristics of the target user can be determined based on the first characteristics of each scene in which the target user is located and the first characteristics of each resource selected by the target user in each scene.

[0082] The resource information includes at least one attribute information, and the first characteristic of the resource can be expressed as i=h(e1, e2, ..., e F ), where i represents the first feature of the resource, h represents the aggregation function, e1, e2, ..., eF They respectively represent the features corresponding to the attribute information, and F represents the amount of attribute information. The features corresponding to the attribute information can be determined according to the formula (2) mentioned above.

[0083] In the embodiment of the present application, the fourth feature of the target user is determined based on the first features of each scene in which the target user is located and the first features of each resource selected by the target user in each scene according to the formula (5) shown below.

[0084] u si =AVG({i j ⊙s j |j=1,2,…,n}) Formula (5)

[0085] Among them, u si The fourth feature that characterizes the target user, AVG, characterizes the average value function, i j The first feature that characterizes the resource selected by the target user in the jth scene, s j The first feature that represents the target user in the jth scene, n is the number of scenes where the target user is located, and ⊙ represents the operator symbol of the Hadamard product.

[0086] After determining the first and fourth characteristics of the target user, the first and fourth characteristics of the target user can be fused to obtain the user characteristics of the target user. This embodiment of the application does not limit the fusion method. For example, the fusion method can be addition, splicing, etc.

[0087] It is understandable that the method for determining the user characteristics of the target user may also vary depending on the application scenario. For example, in one application scenario, the first characteristic of the target user can be directly used as the user characteristic of the target user. In another application scenario, the user characteristics of the target user can be determined based on the second characteristic of the target user and the third characteristic of the target user. Therefore, the embodiment of the present application does not limit the method for determining the user characteristics of the target user, and the user characteristics of the target user can be determined based on at least one of the first characteristic of the target user, the second characteristic of the target user, the third characteristic of the target user, and the fourth characteristic of the target object.

[0088] Optionally, after determining the first feature, the second feature, the third feature, and the fourth feature of the target user, the second feature, the third feature, and the fourth feature of the target user are first concatenated. Then, a multi-layer perceptron is used to fuse the concatenated features, and the fused features are added to the first feature of the target user to obtain the user features of the target user. Among them, u′ represents the user features of the target user, u represents the first feature of the target user, MLP represents the multi-layer perceptron, and u p The second characteristic of the target user is u sp The third feature that characterizes the target user, u si The fourth characteristic that characterizes the target user is Characterize the splicing operation.

[0089] Implementation method B1, the reference information includes object information of multiple objects, and the reference feature is determined based on the reference information, including: for any object, determining the first feature of any object based on the object information of any object; obtaining at least one fourth association information corresponding to any object, and determining the second feature of any object based on at least one fourth association information corresponding to any object, any fourth association information corresponding to any object includes user information of any user when any object is selected by any user; determining the object feature of any object based on the first feature of any object and the second feature of any object.

[0090] In the embodiment of the present application, the heterogeneous scene hypergraph includes object information of multiple objects. For any object, the first feature of the object can be determined based on the object information of the object. The method for determining the first feature of the object has been described in Implementation A1 and will not be repeated here.

[0091] The heterogeneous scene hypergraph in the embodiments of the present application includes a user node, a selection edge, and an object node. For any object, a user node, a selection edge, and any object node can be determined from the heterogeneous scene hypergraph, where any object node is the object node corresponding to any object. The number of user nodes, selection edges, and any object nodes is at least one. For any "user node, selection edge, and any object node," a fourth association information is determined using the user information represented by the user node. In this manner, each fourth association information can be determined.

[0092] In an embodiment of the present application, the first feature of each user when any object is selected by each user can be determined based on each fourth association information, and the second feature of any object can be determined based on the first feature of each user when any object is selected by each user. The method for determining the second feature of any object is similar to the method for determining the second feature of the target user, and will not be repeated here.

[0093] After determining the first feature and the second feature of any object, the first feature and the second feature of any object can be fused to obtain the object feature of any object. The embodiment of the present application does not limit the fusion method. Exemplary fusion methods include addition, splicing, etc.

[0094] Implementation method B2, the reference information includes object information of multiple objects, and the reference features are determined based on the reference information, including: for any object, determining the first feature of any object based on the object information of any object; obtaining at least one fifth association information corresponding to any object, and determining the third feature of any object based on at least one fifth association information corresponding to any object, any fifth association information corresponding to any object includes the scene information of any scene and the user information of any user when any object is selected by any user in any scene; determining the object feature of any object based on the first feature of any object and the third feature of any object.

[0095] For any object, the first feature of the object can be determined based on the object information of the object. The method for determining the first feature of the object has been described in Implementation A1 and will not be repeated here.

[0096] The heterogeneous scene hypergraph in the embodiments of the present application includes user nodes, scene edges, and object nodes. For any object, a user node, scene edge, and any object node can be determined from the heterogeneous scene hypergraph. The number of user nodes, scene edges, and any object nodes is at least one. For any "user node, scene edge, and any object node," a fifth association information is determined using the user information represented by the user node and the scene information represented by the scene edge. In this manner, each fifth association information can be determined.

[0097] In this embodiment of the present application, the first characteristics of each scene in which any object is located and the first characteristics of each user when any object is selected by each user in each scene can be determined based on each fifth association information. The third characteristics of any object can also be determined based on the first characteristics of each scene in which any object is located and the first characteristics of each user when any object is selected by each user in each scene. The third characteristics of any object are determined in a similar manner to the third characteristics of the target user and will not be further described here.

[0098] After determining the first feature and the third feature of any object, the first feature and the third feature of any object can be fused to obtain the object feature of any object. This embodiment of the application does not limit the fusion method. Exemplary fusion methods include addition, splicing, etc.

[0099] Implementation method B3, the reference information includes object information of multiple objects, and the reference characteristics are determined based on the reference information, including: for any object, determining the first characteristic of any object based on the object information of any object; obtaining at least one sixth association information corresponding to any object, and determining the fourth characteristic of any object based on at least one sixth association information corresponding to any object, any sixth association information corresponding to any object includes resource information of any resource published by any object; determining the object characteristic of any object based on the first characteristic of any object and the fourth characteristic of any object.

[0100] For any object, the first feature of the object can be determined based on the object information of the object. The method for determining the first feature of the object has been described in Implementation A1 and will not be repeated here.

[0101] The heterogeneous scenario hypergraph of the embodiment of the present application also includes a publishing edge, and the two ends of the publishing edge are respectively an object node and a resource node, that is, the heterogeneous scenario hypergraph includes an object node-publishing edge-resource node, and the object node-publishing edge-resource node represents that any object publishes any resource. Any object node-publishing edge-resource node can be determined from the heterogeneous scenario hypergraph. The number of any object node-publishing edge-resource nodes is at least one. For any "any object node-publishing edge-resource node", a sixth association information is determined using the resource information represented by the resource node. In this way, each sixth association information can be determined.

[0102] In this embodiment of the present application, the first characteristics of each resource published by any object can be determined based on each sixth association information, and the fourth characteristics of any object can be determined based on the first characteristics of each resource published by any object. The fourth characteristics of any object are determined in a similar manner to the second characteristics of the target user and will not be further described here.

[0103] After determining the first feature and the fourth feature of any object, the first feature and the fourth feature of any object can be fused to obtain the object feature of any object. This embodiment of the application does not limit the fusion method. Exemplary fusion methods include addition, splicing, etc.

[0104] Implementation method B4, the reference information includes object information of multiple objects, and the reference characteristics are determined based on the reference information, including: for any object, determining the first characteristic of any object based on the object information of any object; obtaining at least one seventh related information corresponding to any object, and determining the fifth characteristic of any object based on at least one seventh related information corresponding to any object, any seventh related information corresponding to any object includes the scene information of any scene and the resource information of any resource when any object publishes any resource in any scene; determining the object characteristics of any object based on the first characteristic of any object and the fifth characteristic of any object.

[0105] For any object, the first feature of the object can be determined based on the object information of the object. The method for determining the first feature of the object has been described in Implementation A1 and will not be repeated here.

[0106] The heterogeneous scene hypergraph in the embodiment of the present application includes object nodes, scene edges, and resource nodes. For any object, any object node, scene edge, and resource node can be determined from the heterogeneous scene hypergraph. The number of any object node, scene edge, and resource node is at least one. For any "any object node, scene edge, and resource node," a seventh association information is determined using the scene information represented by the scene edge and the resource information represented by the resource node. In this way, each seventh association information can be determined.

[0107] In this embodiment of the present application, the first characteristics of each scene in which any object resides and the first characteristics of each resource published by any object in each scene can be determined based on each seventh association information. Furthermore, the fifth characteristic of any object can be determined based on the first characteristics of each scene in which any object resides and the first characteristics of each resource published by any object in each scene. The fifth characteristic of any object is determined in a similar manner to the third characteristic of the target user and will not be further described here.

[0108] After determining the first feature and the fifth feature of any object, the first feature and the fifth feature of any object can be fused to obtain the object feature of any object. This embodiment of the application does not limit the fusion method. Exemplary fusion methods include addition, splicing, etc.

[0109] It is understandable that the method for determining the object characteristics of any object may also vary depending on the application scenario. For example, in one application scenario, the first characteristic of any object can be directly used as the object characteristic of any object. Therefore, the embodiment of the present application does not limit the method for determining the object characteristics of any object, and the object characteristics of any object can be determined based on at least one of the first characteristic of any object, the second characteristic of any object, the third characteristic of any object, the fourth characteristic of any object, and the fifth characteristic of any object.

[0110] Optionally, based on the first feature of any object, the second feature of any object, the third feature of any object, the fourth feature of any object, and the fifth feature of any object, the formula To determine the object features of any object. Among them, p′ represents the object features of any object, p represents the first feature of any object, MLP represents the multi-layer perceptron, pu represents the second feature of any object, p su The third characteristic that characterizes any object, p i The fourth characteristic that characterizes any object, p si The fifth characteristic that characterizes any object, Characterize the splicing operation.

[0111] Implementation method Cl, the reference information includes scene information of multiple scenes, and the reference features are determined based on the reference information, including: for any scene, determining the first feature of any scene based on the scene information of any scene; obtaining at least one eighth related information corresponding to any scene, and determining the second feature of any scene based on at least one eighth related information corresponding to any scene, any eighth related information corresponding to any scene includes the scene information of any scene and the user information of any user corresponding to any scene; determining the scene feature of any scene based on the first feature of any scene and the second feature of any scene.

[0112] In this embodiment of the present application, a heterogeneous scene hypergraph includes multiple scene edges, each of which represents the scene information of the scene. For each scene, a first feature of the scene can be determined based on the scene information of the scene. The method for determining the first feature of the scene has been described above and will not be repeated here.

[0113] For any scene edge, the scene edge can be a closed graph containing at least one user node. The user corresponding to the user node contained in the scene edge is the user corresponding to the scene corresponding to the scene edge. Eighth association information is determined using the scene information represented by the scene edge and the user information represented by a user node contained in the scene edge. In this manner, each eighth association information can be determined.

[0114] In an embodiment of the present application, the first feature of any scene and the first feature of each user corresponding to any scene can be determined based on each eighth association information respectively, and the second feature of any scene can be determined based on the first feature of any scene and the first feature of each user corresponding to any scene according to formula (6) shown below.

[0115] s u =AVG({u j ⊙s|j=1,2,…,n}) Formula (6)

[0116] Among them, s u Characterize the second feature of any scene, AVG characterizes the average function, u j represents the first feature of the jth user corresponding to any scene, s represents the first feature of any scene, n is the number of users corresponding to any scene, and ⊙ represents the operator symbol of the Hadamard product.

[0117] After determining the first feature and the second feature of any scene, the first feature and the second feature of any scene can be fused to obtain the scene feature of any scene. The embodiment of the present application does not limit the fusion method. Exemplary fusion methods include addition, splicing, etc.

[0118] Implementation method C2, the reference information includes scene information of multiple scenes, and the reference features are determined based on the reference information, including: for any scene, determining the first feature of any scene based on the scene information of any scene; obtaining at least one ninth associated information corresponding to any scene, and determining the third feature of any scene based on at least one ninth associated information corresponding to any scene, any ninth associated information corresponding to any scene includes the scene information of any scene and the object information of any object corresponding to any scene; determining the scene feature of any scene based on the first feature of any scene and the third feature of any scene.

[0119] In an embodiment of the present application, for any scene, the first feature of the scene can be determined based on the scene information of the scene. The method for determining the first feature of the scene has been described above and will not be repeated here.

[0120] For any scene edge, the scene edge can be a closed graph containing at least one object node. The object corresponding to the object node contained in the scene edge is the object corresponding to the scene corresponding to the scene edge. A ninth piece of association information is determined using the scene information represented by the scene edge and the object information represented by an object node contained in the scene edge. In this manner, each piece of ninth association information can be determined.

[0121] In an embodiment of the present application, the first feature of any scene and the first feature of each object corresponding to any scene can be determined respectively based on each ninth association information, and the third feature of any scene can be determined according to formula (7) shown below based on the first feature of any scene and the first feature of each object corresponding to any scene.

[0122] s p =AVG({p j ⊙s|j=1,2,…,n}) Formula (7)

[0123] Among them, s p Characterize the third feature of any scene, AVG characterizes the average function, p j represents the first feature of the j-th object corresponding to any scene, s represents the first feature of any scene, n is the number of objects corresponding to any scene, and ⊙ represents the operator symbol of the Hadamard product.

[0124] After determining the first feature and the third feature of any scene, the first feature and the third feature of any scene can be fused to obtain the scene feature of any scene. The embodiment of the present application does not limit the fusion method. Exemplary fusion methods include addition, splicing, etc.

[0125] It can be understood that the scene feature of any scene can be determined based on at least one of the first feature of any scene, the second feature of any scene, and the third feature of any scene.

[0126] Optionally, after determining the first feature, the second feature, and the third feature of any scene, the second feature and the third feature of any scene are first concatenated, and then the concatenated features are fused using a multi-layer perceptron, and the fused features are added to the first feature of any scene to obtain the scene feature of any scene. Among them, s′ represents the scene feature of any scene, s represents the first feature of any scene, MLP represents the multi-layer perceptron, s u The second characteristic of any scene, s p The third characteristic that characterizes any scene, Characterize the splicing operation.

[0127] Step 203: Determine an object to be recommended from multiple objects based on the reference features.

[0128] In an embodiment of the present application, an algorithm or model can be used to determine an object to be recommended from multiple objects based on reference features. That is, an algorithm or model can be used to determine an object to be recommended from multiple objects based on the scene features of each scene. Alternatively, an algorithm or model can be used to determine an object to be recommended from multiple objects based on the user features of the target user, at least one of the object features of each object, and the scene features of each scene. Alternatively, an algorithm or model can be used to determine an object to be recommended from multiple objects based on the user features of the target user, the object features of each object, and the scene features of each scene.

[0129] In one possible implementation, based on reference features, objects to be recommended are determined from multiple objects, including: based on the reference features, indicator information of each object is determined, where the indicator information of any object is used to characterize the degree of matching between the target user and any object; based on the indicator information of each object, objects to be recommended whose indicator information meets the filtering conditions are screened out from multiple objects.

[0130] In an embodiment of the present application, the index information of each object can be determined based on the scene characteristics of each scene, or based on the user characteristics of the target user, at least one of the object characteristics of each object, and the scene characteristics of each scene. The index information of any object can be a probability greater than or equal to 0 and less than or equal to 1, or it can be 0 or a positive number. Among them, the larger the index information of any object, the higher the degree of match between the target user and any object.

[0131] For any object, if the object's indicator information is greater than the first threshold, it indicates that the object's indicator information meets the screening conditions and the object can be recommended. If the object's indicator information is not greater than the first threshold, it indicates that the object's indicator information does not meet the screening conditions and the object cannot be recommended. In this way, objects to be recommended can be screened out from multiple objects.

[0132] In an embodiment of the present application, it is possible to first determine a scene to be recommended from multiple scenes based on the scene characteristics of each scene, and then determine an object to be recommended from multiple objects based on the scene characteristics of the scene to be recommended. It is also possible to first determine a scene to be recommended from multiple scenes based on the user characteristics of the target user and the scene characteristics of each scene, and then determine an object to be recommended from multiple objects based on the scene characteristics of the scene to be recommended. It is also possible to first determine a scene to be recommended from multiple scenes based on the scene characteristics of each scene, and then determine an object to be recommended from multiple objects based on the scene characteristics of the scene to be recommended and the object characteristics of each object.

[0133] In one possible implementation, the reference features include user features of the target user, object features of each object, and scene features of each scene. Based on the reference features, the object to be recommended is determined from multiple objects, including: determining the scene to be recommended from multiple scenes based on the user features of the target user and the scene features of each scene; determining the object to be recommended from multiple objects based on the scene features of the scene to be recommended and the object features of each object.

[0134] In the embodiment of the present application, the indicator information of each scenario is first determined based on the user characteristics of the target user and the scene characteristics of each scenario. The indicator information of any scenario represents the degree of match between the target user and any scenario. The indicator information of any scenario can be a probability greater than or equal to 0 and less than or equal to 1, or it can be 0 or a positive number. Among them, the larger the indicator information of any scenario, the higher the degree of match between the target user and any scenario.

[0135] For any scene, if the scene's indicator information is greater than the second threshold, it indicates that the scene's indicator information meets the target conditions and the scene can be selected as a recommended scene. If the scene's indicator information is not greater than the second threshold, it indicates that the scene's indicator information does not meet the target conditions and the scene cannot be selected as a recommended scene. In this way, scenes to be recommended can be screened from multiple scenes.

[0136] Optionally, based on the user characteristics of the target user and the scene characteristics of each scene, the scene to be recommended is determined from multiple scenes, including: obtaining environmental information of multiple environments, and determining the environmental characteristics of each environment based on the environmental information of each environment; based on the user characteristics of the target user, the scene characteristics of each scene and the environmental characteristics of each environment, determining the scene to be recommended from multiple scenes.

[0137] The heterogeneous scenario hypergraph in the embodiment of the present application also includes multiple environment edges, each representing the environmental information of an environment, including at least one of attribute information such as time, location, and weather. The environment edge can be a closed graph that includes user nodes, object nodes, and resource nodes.

[0138] It is understood that since both environment edges and scene edges can be closed graphs containing user nodes, object nodes, and resource nodes, to simplify the heterogeneous scene hypergraph, the same closed graph can be used to represent both environment edges and scene edges. For example, the closed graph itself is the environment edge, and the color of the closed graph is the scene edge.

[0139] For any environment, the first characteristic of the environment can be determined based on the environment information of the environment, and the environment characteristic of the environment can be determined based on the first characteristic of the environment. Since the environment information includes at least one attribute information, the first characteristic of the environment can be expressed as c=h(e1, e2, ..., e F ), where c represents the first characteristic of the environment, h represents the aggregation function, e1, e2, ..., e F They respectively represent the features corresponding to the attribute information, and F represents the amount of attribute information. The features corresponding to the attribute information can be determined according to the formula (2) mentioned above.

[0140] In the embodiment of the present application, the indicator information of each scene can be determined according to the following formula (8) based on the user characteristics of the target user, the scene characteristics of each scene, and the environmental characteristics of each environment.

[0141] g(s|c,u)=sum(c⊙u⊙s) Formula (8)

[0142] Among them, g(s|c,u) represents the indicator information of a scene, sum represents the summation function, c represents the environmental characteristics of an environment, u represents the user characteristics of the target user, s represents the scene characteristics of a scene, and ⊙ represents the operator symbol of the Hadamard product.

[0143] After determining the indicator information of each scene, based on the indicator information of each scene, a scene to be recommended is determined from the multiple scenes. Optionally, the indicator information of each scene is sorted from largest to smallest, and the top several scenes are selected as the scenes to be recommended according to formula (9) shown below.

[0144]

[0145] in, represents the scene to be recommended, j is the sequence number, k s Represents the number of scenes to be recommended, Representation of the indicator information of each scene is sorted from large to small, and the top k s scenes as the recommended scenes.

[0146] After the scene to be recommended is determined, based on the scene features of the scene to be recommended and the object features of each object, the index information of each object in the scene to be recommended is determined according to the following formula (10).

[0147]

[0148] in, Represents the index information of an object in the scene to be recommended, sum represents the summation function, c represents the environmental characteristics of an environment, u represents the user characteristics of the target user, s represents the scene characteristics of a scene, p represents the object characteristics of an object, and ⊙ represents the operator symbol of the Hadamard product. Among them, formula (8) and formula (9) are reused in formula (10), and formula (8) and formula (9) can be used to determine the scene to be recommended.

[0149] After determining the index information of each object in the scenario to be recommended, an object to be recommended is determined from the plurality of objects based on the index information of each object in the scenario to be recommended. Optionally, the index information of each object in the scenario to be recommended is sorted from largest to smallest, and the top several objects in the sorting are selected as the objects to be recommended according to formula (11) shown below.

[0150]

[0151] in, represents the object to be recommended, j is the sequence number, k p Represents the number of objects to be recommended, Representation of the set of scenes to be recommended Any of the recommended scenarios Characterize the scene to be recommended Sort the indicator information of each object from large to small, and select the top k p objects as the recommended objects.

[0152] In one possible implementation, determining reference features based on reference information includes: obtaining a recommendation model, the recommendation model including a feature extraction sub-model and a feature processing sub-model connected in sequence; determining the reference features by the feature extraction sub-model based on the reference information; and determining an object to be recommended from multiple objects based on the reference features, including: determining the object to be recommended from multiple objects by the feature processing sub-model based on the reference features.

[0153] A recommendation model can be used to determine at least one object to be recommended from multiple objects based on the scene features of each scene, or based on the user features of the target user, at least one of the object features of each object, and the scene features of each scene. The recommendation model includes a feature extraction submodel and a feature processing submodel connected in sequence. This embodiment of the application does not limit the model structure and model size of the recommendation model.

[0154] Optionally, a heterogeneous scene hypergraph is input into the recommendation model, and the feature extraction sub-model determines the target user's user features based on the target user's user information, determines the object features of each object based on the object information of each object, and determines the scene features of each scene based on the scene information of each scene. The feature processing sub-model determines the index information of each object based on the user features of the target user, the object features of each object, and the scene features of each scene. Based on the index information of each object, the recommended object is screened from multiple objects to select the object whose index information meets the screening criteria.

[0155] Among them, the recommendation model can be obtained based on the training of the neural network model. Optionally, a sample heterogeneous scene hypergraph is obtained, and the sample heterogeneous scene hypergraph can be recorded as the heterogeneous scene hypergraph of the positive sample. The heterogeneous scene hypergraph of the positive sample includes user information of multiple sample users, object information of multiple sample objects, and scene information of multiple sample scenes. At least one of the sample users, sample objects, sample scenes, etc. in the heterogeneous scene hypergraph of the positive sample is replaced to obtain the heterogeneous scene hypergraph of the negative sample. For the content of the sample heterogeneous scene hypergraph, please refer to the description of the heterogeneous scene hypergraph above. The implementation principles of the two are similar and will not be repeated here.

[0156] In the embodiment of the present application, a neural network model is used to determine the index information of each sample object corresponding to the positive sample based on the heterogeneous scene hypergraph of the positive sample, and a neural network model is used to determine the index information of each sample object corresponding to the negative sample based on the heterogeneous scene hypergraph of the negative sample. The index information of any sample object is used to characterize the degree of match between the target sample user (one sample user among multiple sample users) and any sample object. The method for determining the index information of each sample object is described in steps 202 and 203 and will not be repeated here.

[0157] Next, according to the following formula (12), the loss value of the neural network model is determined using the indicator information of each sample object corresponding to the positive and negative samples. After that, the neural network model is trained based on the loss value of the neural network model to obtain the recommendation model.

[0158]

[0159]

[0160] Among them, (f(p|c,u;s)) represents the index information of any sample object corresponding to the positive sample or negative sample, sigmoid represents the activation function, Characterizes the probability of any sample object corresponding to a positive sample or a negative sample. Characterize the loss value of the neural network model, Characterize the number of sample objects corresponding to the positive samples, Characterizes the number of sample objects corresponding to negative samples. j Represents the annotation information of the j-th sample object, Characterize the probability of the j-th sample object.

[0161] Step 204: recommend the object to be recommended to the target user.

[0162] In the embodiment of the present application, k can be determined s scenes to be recommended, for any scene to be recommended, k p objects to be recommended. Therefore, k s ·k p Recommend the recommended objects to the target users.

[0163] The above describes the recommended method of the embodiment of the present application from the perspective of method steps. Figure 5 and Figure 6 , further illustrating the recommended method of the embodiments of this application.

[0164] See Figure 5 , Figure 5 Schematic diagram of a heterogeneous scenario hypergraph provided in an embodiment of the present application. The heterogeneous scenario hypergraph in this embodiment includes three types of nodes: user nodes, object nodes, and resource nodes. It also includes three types of edges: selection edges, release edges, environment edges, and scenario edges. Selection edges are represented by thick solid lines, release edges by dashed lines, and environment edges and scenario edges are the same type of edge and are represented by closed graphs.

[0165] The closed graph corresponding to the environment edge c1 and the scene edge s1 includes the user node u1, the object node p1, the resource node i1, the resource node i2, the selection edge between the user node u1 and the object node p1, the release edge between the object node p1 and the resource node i1, and the release edge between the object node p1 and the resource node i2. The closed graph corresponding to the environment edge c2 and the scene edge s2 includes the user node u2, the object node p1, the resource node i3, the selection edge between the user node u2 and the object node p1, and the release edge between the object node p1 and the resource node i3. The closed graph corresponding to the environment edge c3 and the scene edge s3 includes the user node u1, the object node p2, the resource node i4, the resource node i5, the release edge between the object node p2 and the resource node i4, and the release edge between the object node p2 and the resource node i5.

[0166] Next, see Figure 6 , Figure 6This is a schematic diagram of determining an object to be recommended, provided in an embodiment of the present application. For environment c, the features corresponding to each attribute information of environment c are obtained by looking up the coding table. By normalizing the features corresponding to each attribute information of environment c, the features corresponding to environment c are obtained. These features corresponding to environment c are the first features of environment c mentioned above.

[0167] Based on the same principle, for user u, by looking up the coding table and normalizing, we can obtain the features corresponding to user u, which is the first feature of user u mentioned above. For scene s, by looking up the coding table and normalizing, we can obtain the features corresponding to scene s, which is the first feature of scene s mentioned above. For object p, by looking up the coding table and normalizing, we can obtain the features corresponding to object p, which is the first feature of object p mentioned above. For resource i, by looking up the coding table and normalizing, we can obtain the features corresponding to resource i, which is the first feature of resource i mentioned above.

[0168] For environment c, a feature corresponding to another environment c may be determined based on the first feature of environment c. The feature corresponding to the environment c is the environmental feature of environment c mentioned above.

[0169] For user u, by multiplying the first feature of object p and the first feature of scene s, a feature corresponding to user u can be determined. The feature corresponding to user u is the third feature of user u mentioned above. Based on the first feature of object p, a feature corresponding to another user u can be determined. The feature corresponding to user u is the second feature of user u mentioned above. By multiplying the first feature of resource i and the first feature of scene s, a feature corresponding to yet another user u can be determined. The feature corresponding to user u is the fourth feature of user u mentioned above. Then, after the second feature of user u, the third feature of user u, and the fourth feature of user u are fused through a multi-layer perceptron, the fused features are merged with the first feature of user u to obtain a feature corresponding to user u. The feature corresponding to user u is the user feature of user u mentioned above.

[0170] For scene s, multiplying the first feature of user u with the first feature of scene s determines a feature corresponding to scene s. This feature is the second feature of scene s mentioned above. Multiplying the first feature of object p with the first feature of scene s determines another feature corresponding to scene s. This feature is the third feature of scene s mentioned above. Then, after fusing the second and third features of scene s using a multilayer perceptron, the fused feature is combined with the first feature of scene s to obtain a feature corresponding to scene s. This feature is the scene feature of scene s mentioned above.

[0171] For object p, based on the first feature of user u, a feature corresponding to object p can be determined. This feature corresponding to object p is the second feature of object p mentioned above. Based on the first feature of resource i, another feature corresponding to object p can be determined. This feature corresponding to object p is the fourth feature of object p mentioned above. By multiplying the first feature of user u and the first feature of scene s, a feature corresponding to yet another object p can be determined. This feature corresponding to object p is the third feature of object p mentioned above. By multiplying the first feature of resource i and the first feature of scene s, a feature corresponding to yet another object p can be determined. This feature corresponding to object p is the fifth feature of object p mentioned above. Then, after the second feature of object p, the third feature of object p, the fourth feature of object p, and the fifth feature of object p are fused through a multi-layer perceptron, the fused features are merged with the first feature of object p to obtain a feature corresponding to object p. This feature corresponding to object p is the object feature of object p mentioned above.

[0172] Next, based on the environmental characteristics of environment c, the user characteristics of user u, and the scene characteristics of scene s, a scene s to be recommended is determined from each scene s, and the scene characteristics of the scene s to be recommended are determined. Then, based on the scene characteristics of the scene s to be recommended and the object characteristics of the object p, an object p to be recommended is determined from each object p. The method for determining the scenes and objects to be recommended is described above in relation to step 203 and will not be repeated here.

[0173] This example captures eight days of interaction behavior. For each interaction, a positive sample is generated. A positive sample consists of a user, an object, a scene, and a triplet of "scene-object-user." A negative sample is generated by replacing at least one of the user, object, or scene in a positive sample.

[0174] The embodiment of the present application uses 8 days of positive and negative samples to construct 4 data sets. These 4 data sets are recorded as 1 day, 3 days, 5 days and 7 days respectively. Each data set includes a training set and a test set. The training set corresponding to n days (n is 1, 3, 5 or 7) is constructed using the positive and negative samples from the 1st to nth days, and the test set corresponding to the nth day is constructed using the positive and negative samples of the n+1th day. For example, the training set corresponding to 5 days is constructed using the positive and negative samples from the 1st to 5th days, and the test set corresponding to 5 days is constructed using the positive and negative samples of the 6th day.

[0175] The number of positive samples, negative samples, users, objects, scene-object-users, and scenes in each training set and test set were counted. In addition, the number of new users, new objects, new scene-object-users, and new scenes in the test set were counted compared with the training set, resulting in Table 1 as shown below.

[0176] Table 1

[0177]

[0178] For each training set in Table 1, 50%, 75%, and 100% of the training set are used to train the recommendation model, and the recommendation model is used to output the recommended objects for each user in the test set. The output results of the recommendation model and the test set are used to calculate the area under the curve (AUC), and the AUC value corresponding to the object recommendation method of the embodiment of the present application (referred to as this method) is obtained. At the same time, the embodiment of the present application also calculates the AUC value of other methods such as Deep Factorization Machine (DeepFM), DeepFM S , Automatic Feature Interaction Learning Via Self-Attentive Neural Networks (AutoInt), AutoInt S , Neighborhood-based Interaction Model ForRecommendation (NIRec), NIRec S , Heterogeneous Graph Network - Intention Recommendation (MEIRec), Hierarchical Attention Networks (HAN), Heterogeneous Graph Attention Network (HGAT), A Self-attention Based Graph Neural Network For Hypergraphs (Hyper-SAGNN), Hyper-SAGNN S The corresponding AUC values ​​are obtained by comparing this method with other methods to obtain the improvement of this method (in %), as shown in Table 2 below.

[0179] Table 2

[0180]

[0181]

[0182] As can be seen from Table 2, compared with other methods, this method has an improvement in AUC value, indicating that the recommended method of the embodiment of the present application can improve accuracy.

[0183] In addition, this application also calculates the recall rate indicators of this method, HAN, and HGAT. The recall rate indicator can be the HR@K indicator. Taking the HR@K indicators of HR@10-S and HR@100-P as examples, the following Table 3 is obtained.

[0184] Table 3

[0185]

[0186] As shown in Table 3, compared with HAN and HGAT, the HR@10-S and HR@100-P of the present method are larger, indicating that the method of the embodiment of the present application has higher accuracy.

[0187] The above method determines the object to be recommended from multiple objects based on the scene characteristics of each scene, or based on the user characteristics of the target user, at least one of the object characteristics of each object and the scene characteristics of each scene, thereby realizing the determination of the object to be recommended based on scene information or comprehensive user information, at least one of the object information and scene information, improving accuracy, and thus increasing the time and frequency of users using the application.

[0188] Figure 7 FIG. 1 is a schematic diagram of the structure of an object recommendation device provided in an embodiment of the present application. Figure 7 As shown, the device includes:

[0189] An acquisition module 701 is configured to acquire reference information, where the reference information includes scenario information of multiple scenarios, or the reference information includes user information of a target user, at least one of object information of multiple objects, and scenario information of multiple scenarios, where the scenario represents a type of interactive behavior, where the interactive behavior is the behavior of a user selecting resources published by an object in an environment;

[0190] A determination module 702 is configured to determine reference features based on the reference information, where the reference features include scene features of each scene, or the reference features include user features of a target user, at least one of object features of each object, and scene features of each scene;

[0191] The determination module 702 is further configured to determine an object to be recommended from a plurality of objects based on the reference features;

[0192] The recommendation module 703 is used to recommend the object to be recommended to the target user.

[0193] In one possible implementation, the reference information includes user information of the target user, and the determination module 702 is used to determine a first feature of the target user based on the user information of the target user; obtain at least one first association information, and determine a second feature of the target user based on the at least one first association information, where any first association information includes object information of any object selected by the target user; and determine the user feature of the target user based on the first feature of the target user and the second feature of the target user.

[0194] In one possible implementation, the reference information includes user information of the target user, and the determination module 702 is used to determine a first feature of the target user based on the user information of the target user; obtain at least one second association information, and determine a third feature of the target user based on the at least one second association information, where any second association information includes scene information of any scene and object information of any object when the target user selects any object in any scene; and determine the user feature of the target user based on the first feature of the target user and the third feature of the target user.

[0195] In one possible implementation, the reference information includes user information of the target user, and the determination module 702 is used to determine a first feature of the target user based on the user information of the target user; obtain at least one third association information, and determine a fourth feature of the target user based on the at least one third association information, where any third association information includes scene information of any scene and resource information of any resource when the target user selects any resource in any scene; and determine the user feature of the target user based on the first feature of the target user and the fourth feature of the target user.

[0196] In one possible implementation, the reference information includes object information of multiple objects, and the determination module 702 is used to determine, for any object, a first feature of any object based on the object information of any object; obtain at least one fourth association information corresponding to any object, and determine a second feature of any object based on at least one fourth association information corresponding to any object, and any fourth association information corresponding to any object includes user information of any user when any object is selected by any user; and determine the object feature of any object based on the first feature of any object and the second feature of any object.

[0197] In one possible implementation, the reference information includes object information of multiple objects, and the determination module 702 is used to determine, for any object, a first feature of any object based on the object information of any object; obtain at least one fifth association information corresponding to any object, and determine a third feature of any object based on at least one fifth association information corresponding to any object, and any fifth association information corresponding to any object includes scene information of any scene and user information of any user when any object is selected by any user in any scene; and determine the object feature of any object based on the first feature of any object and the third feature of any object.

[0198] In one possible implementation, the reference information includes object information of multiple objects, and the determination module 702 is used to determine, for any object, a first feature of any object based on the object information of any object; obtain at least one sixth association information corresponding to any object, and determine a fourth feature of any object based on at least one sixth association information corresponding to any object, and any sixth association information corresponding to any object includes resource information of any resource published by any object; and determine the object feature of any object based on the first feature of any object and the fourth feature of any object.

[0199] In one possible implementation, the reference information includes object information of multiple objects, and the determination module 702 is used to determine the first feature of any object based on the object information of any object; obtain at least one seventh association information corresponding to any object, and determine the fifth feature of any object based on the at least one seventh association information corresponding to any object, and any seventh association information corresponding to any object includes the scene information of any scene and the resource information of any resource when any object publishes any resource in any scene; determine the object feature of any object based on the first feature of any object and the fifth feature of any object.

[0200] In one possible implementation, the reference information includes scene information of multiple scenes, and the determination module 702 is used to determine, for any scene, a first feature of any scene based on the scene information of any scene; obtain at least one eighth associated information corresponding to any scene, and determine a second feature of any scene based on at least one eighth associated information corresponding to any scene, and any eighth associated information corresponding to any scene includes the scene information of any scene and user information of any user corresponding to any scene; and determine the scene feature of any scene based on the first feature of any scene and the second feature of any scene.

[0201] In one possible implementation, the reference information includes scene information of multiple scenes, and the determination module 702 is used to determine, for any scene, the first feature of any scene based on the scene information of any scene; obtain at least one ninth associated information corresponding to any scene, and determine the third feature of any scene based on at least one ninth associated information corresponding to any scene, and any ninth associated information corresponding to any scene includes the scene information of any scene and the object information of any object corresponding to any scene; and determine the scene feature of any scene based on the first feature of any scene and the third feature of any scene.

[0202] In one possible implementation, the reference features include user features of the target user, object features of each object, and scene features of each scene; the determination module 702 is used to determine the scene to be recommended from multiple scenes based on the user features of the target user and the scene features of each scene; and to determine the object to be recommended from multiple objects based on the scene features of the scene to be recommended and the object features of each object.

[0203] In one possible implementation, the determination module 702 is used to obtain environmental information of multiple environments, and determine the environmental characteristics of each environment based on the environmental information of each environment; based on the user characteristics of the target user, the scene characteristics of each scene, and the environmental characteristics of each environment, determine the scene to be recommended from the multiple scenes.

[0204] In one possible implementation, the determination module 702 is used to determine the indicator information of each object based on the reference features, where the indicator information of any object is used to characterize the degree of matching between the target user and any object; based on the indicator information of each object, objects to be recommended whose indicator information meets the filtering conditions are filtered out from multiple objects.

[0205] In one possible implementation, the determination module 702 is used to obtain a recommendation model, which includes a feature extraction sub-model and a feature processing sub-model connected in sequence; the feature extraction sub-model determines the reference feature based on the reference information; and the feature processing sub-model determines the object to be recommended from multiple objects based on the reference feature.

[0206] The above-mentioned device determines the object to be recommended from multiple objects based on the scene characteristics of each scene, or based on the user characteristics of the target user, at least one of the object characteristics of each object and the scene characteristics of each scene, and realizes the determination of the object to be recommended based on scene information or comprehensive user information, at least one of the object information and scene information, thereby improving accuracy and thus increasing the time and frequency of users using the application.

[0207] It should be understood that the above Figure 7The provided device is illustrated only by the division of the above-mentioned functional modules when implementing its functions. In actual applications, the above-mentioned functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. In addition, the device and method embodiments provided in the above embodiments are based on the same concept. The specific implementation process is detailed in the method embodiment and will not be repeated here.

[0208] Figure 8 The following is a block diagram of a terminal device 800 according to an exemplary embodiment of the present application. The terminal device 800 may be a portable mobile terminal, such as a smartphone, tablet computer, MP3 (Moving Picture Experts Group Audio Layer III) player, MP4 (Moving Picture Experts Group Audio Layer IV) player, laptop computer, or desktop computer. The terminal device 800 may also be referred to as a user device, portable terminal, laptop terminal, desktop terminal, or other similar terminology.

[0209] Typically, the terminal device 800 includes a processor 801 and a memory 802 .

[0210] The processor 801 may include one or more processing cores, such as a 4-core processor, an 8-core processor, etc. The processor 801 may be implemented in at least one hardware form of DSP (Digital Signal Processing), FPGA (Field-Programmable Gate Array), or PLA (Programmable Logic Array). The processor 801 may also include a main processor and a coprocessor. The main processor is a processor for processing data in the awake state, also known as a CPU (Central Processing Unit); the coprocessor is a low-power processor for processing data in the standby state. In some embodiments, the processor 801 may be integrated with a GPU (Graphics Processing Unit), which is responsible for rendering and drawing the content to be displayed on the display screen. In some embodiments, the processor 801 may also include an AI (Artificial Intelligence) processor, which is used to process computing operations related to machine learning.

[0211] The memory 802 may include one or more computer-readable storage media, which may be non-transitory. The memory 802 may also include high-speed random access memory and non-volatile memory, such as one or more disk storage devices and flash memory storage devices. In some embodiments, the non-transitory computer-readable storage medium in the memory 802 is used to store at least one instruction, which is used to be executed by the processor 801 to implement the object recommendation method provided in the method embodiment of the present application.

[0212] In some embodiments, terminal device 800 may also optionally include a peripheral device interface 803 and at least one peripheral device. The processor 801, memory 802, and peripheral device interface 803 may be connected via a bus or signal lines. Each peripheral device may be connected to peripheral device interface 803 via a bus, signal lines, or circuit boards. Specifically, the peripheral device may include at least one of a radio frequency circuit 804, a display screen 805, a camera assembly 806, an audio circuit 807, a positioning assembly 808, and a power supply 809.

[0213] The peripheral device interface 803 can be used to connect at least one I / O (Input / Output)-related peripheral device to the processor 801 and the memory 802. In some embodiments, the processor 801, the memory 802, and the peripheral device interface 803 are integrated on the same chip or circuit board; in some other embodiments, any one or two of the processor 801, the memory 802, and the peripheral device interface 803 can be implemented on separate chips or circuit boards, which is not limited in this embodiment.

[0214] The radio frequency circuit 804 is used to receive and transmit RF (Radio Frequency) signals, also known as electromagnetic signals. The radio frequency circuit 804 communicates with communication networks and other communication devices via electromagnetic signals. The radio frequency circuit 804 converts electrical signals into electromagnetic signals for transmission, or converts received electromagnetic signals into electrical signals. Optionally, the radio frequency circuit 804 includes: an antenna system, an RF transceiver, one or more amplifiers, a tuner, an oscillator, a digital signal processor, a codec chipset, a user identity module card, etc. The radio frequency circuit 804 can communicate with other terminals via at least one wireless communication protocol. Such wireless communication protocols include, but are not limited to, the World Wide Web, a metropolitan area network, an intranet, various generations of mobile communication networks (2G, 3G, 4G, and 5G), a wireless local area network, and / or a WiFi (Wireless Fidelity) network. In some embodiments, the radio frequency circuit 804 may also include circuits related to NFC (Near Field Communication), which is not limited in this application.

[0215] The display screen 805 is used to display a user interface (UI). This UI may include graphics, text, icons, videos, or any combination thereof. When the display screen 805 is a touch screen, it is also capable of collecting touch signals on or above the surface of the display screen 805. These touch signals can be input as control signals to the processor 801 for processing. In this case, the display screen 805 can also be used to provide virtual buttons and / or a virtual keyboard, also known as soft buttons and / or a soft keyboard. In some embodiments, there can be one display screen 805, located on the front panel of the terminal device 800. In other embodiments, there can be at least two display screens 805, located on different surfaces of the terminal device 800 or in a foldable design. In other embodiments, the display screen 805 can be a flexible display, located on a curved or foldable surface of the terminal device 800. Furthermore, the display screen 805 can be configured as a non-rectangular irregular shape, i.e., a special-shaped screen. The display screen 805 can be made of materials such as LCD (Liquid Crystal Display) and OLED (Organic Light-Emitting Diode).

[0216] The camera assembly 806 is used to capture images or videos. Optionally, the camera assembly 806 includes a front camera and a rear camera. Typically, the front camera is arranged on the front panel of the terminal, and the rear camera is arranged on the back of the terminal. In some embodiments, there are at least two rear cameras, which are any one of a main camera, a depth of field camera, a wide-angle camera, and a telephoto camera, so as to realize the fusion of the main camera and the depth of field camera to realize the background blur function, the fusion of the main camera and the wide-angle camera to realize panoramic shooting and VR (Virtual Reality) shooting function or other fusion shooting functions. In some embodiments, the camera assembly 806 may also include a flash. The flash can be a monochrome temperature flash or a dual-color temperature flash. A dual-color temperature flash refers to a combination of a warm light flash and a cold light flash, which can be used for light compensation at different color temperatures.

[0217] The audio circuit 807 may include a microphone and a speaker. The microphone is used to collect sound waves from the user and the environment, and convert the sound waves into electrical signals that are input into the processor 801 for processing, or input into the radio frequency circuit 804 to achieve voice communication. For the purpose of stereo sound collection or noise reduction, there can be multiple microphones, each located in different parts of the terminal device 800. The microphone can also be an array microphone or an omnidirectional collection microphone. The speaker is used to convert electrical signals from the processor 801 or the radio frequency circuit 804 into sound waves. The speaker can be a traditional thin film speaker or a piezoelectric ceramic speaker. When the speaker is a piezoelectric ceramic speaker, it can not only convert electrical signals into sound waves audible to humans, but also convert electrical signals into sound waves inaudible to humans for purposes such as ranging. In some embodiments, the audio circuit 807 may also include a headphone jack.

[0218] The positioning component 808 is used to locate the current geographic location of the terminal device 800 to implement navigation or LBS (Location Based Service). The positioning component 808 can be a positioning component based on the US GPS (Global Positioning System), China's Beidou system, or Russia's Galileo system.

[0219] Power supply 809 is used to power the various components in terminal device 800. Power supply 809 can be AC ​​power, DC power, a disposable battery, or a rechargeable battery. When power supply 809 includes a rechargeable battery, the rechargeable battery can be a wired rechargeable battery or a wireless rechargeable battery. A wired rechargeable battery is a battery that is charged via a wired line, while a wireless rechargeable battery is a battery that is charged via a wireless coil. The rechargeable battery can also be used to support fast charging technology.

[0220] In some embodiments, the terminal device 800 further includes one or more sensors 810 , including but not limited to: an acceleration sensor 811 , a gyroscope sensor 812 , a pressure sensor 813 , an optical sensor 814 , and a proximity sensor 815 .

[0221] The accelerometer 811 can detect the magnitude of acceleration along the three coordinate axes of the coordinate system established by the terminal device 800. For example, the accelerometer 811 can be used to detect the components of gravity acceleration along the three coordinate axes. The processor 801 can control the display screen 805 to display the user interface in a landscape or portrait view based on the gravity acceleration signal collected by the accelerometer 811. The accelerometer 811 can also be used to collect game or user motion data.

[0222] The gyroscope sensor 812 can detect the orientation and rotation angle of the terminal device 800. The gyroscope sensor 812 can work with the accelerometer 811 to collect the user's 3D movements of the terminal device 800. Based on the data collected by the gyroscope sensor 812, the processor 801 can implement the following functions: motion sensing (such as changing the UI based on the user's tilt operation), image stabilization during shooting, game control, and inertial navigation.

[0223] The pressure sensor 813 can be set on the side frame of the terminal device 800 and / or the lower layer of the display screen 805. When the pressure sensor 813 is set on the side frame of the terminal device 800, it can detect the user's grip signal of the terminal device 800, and the processor 801 performs left and right hand recognition or shortcut operations based on the grip signal collected by the pressure sensor 813. When the pressure sensor 813 is set on the lower layer of the display screen 805, the processor 801 controls the operable controls on the UI interface based on the user's pressure operation on the display screen 805. The operable controls include at least one of a button control, a scroll bar control, an icon control, and a menu control.

[0224] Optical sensor 814 is used to detect ambient light intensity. In one embodiment, processor 801 can control the display brightness of display screen 805 based on the ambient light intensity detected by optical sensor 814. Specifically, when the ambient light intensity is high, the display brightness of display screen 805 is increased; when the ambient light intensity is low, the display brightness of display screen 805 is decreased. In another embodiment, processor 801 can also dynamically adjust the shooting parameters of camera assembly 806 based on the ambient light intensity detected by optical sensor 814.

[0225] The proximity sensor 815, also known as a distance sensor, is typically located on the front panel of the terminal device 800. The proximity sensor 815 is used to detect the distance between the user and the front of the terminal device 800. In one embodiment, when the proximity sensor 815 detects that the distance between the user and the front of the terminal device 800 is gradually decreasing, the processor 801 controls the display screen 805 to switch from the screen-on state to the screen-off state. When the proximity sensor 815 detects that the distance between the user and the front of the terminal device 800 is gradually increasing, the processor 801 controls the display screen 805 to switch from the screen-off state to the screen-on state.

[0226] Those skilled in the art will understand that Figure 8 The structure shown in the figure does not constitute a limitation on the terminal device 800, and the terminal device 800 may include more or fewer components than shown in the figure, or combine certain components, or adopt a different component arrangement.

[0227] Figure 9This is a schematic diagram of the structure of the server provided in an embodiment of the present application. The server 900 may have relatively large differences due to different configurations or performances, and may include one or more processors 901 and one or more memories 902, wherein the one or more memories 902 store at least one program code, and the at least one program code is loaded and executed by the one or more processors 901 to implement the object recommendation method provided by the above-mentioned various method embodiments. Exemplarily, the processor 901 is a CPU. Of course, the server 900 may also have components such as a wired or wireless network interface, a keyboard, and an input and output interface for input and output. The server 900 may also include other components for implementing device functions, which will not be described in detail here.

[0228] In an exemplary embodiment, a computer-readable storage medium is further provided. The storage medium stores at least one program code. The at least one program code is loaded and executed by a processor to enable an electronic device to implement any of the above-mentioned object recommendation methods.

[0229] Optionally, the computer-readable storage medium may be a read-only memory (ROM), a random access memory (RAM), a compact disc (CD-ROM), a magnetic tape, a floppy disk, an optical data storage device, or the like.

[0230] In an exemplary embodiment, a computer program or a computer program product is further provided. The computer program or the computer program product stores at least one computer instruction, and the at least one computer instruction is loaded and executed by a processor to enable a computer to implement any of the above-mentioned object recommendation methods.

[0231] It should be understood that the term "plurality" used herein refers to two or more. "And / or" describes a relationship between associated objects, indicating that three possible relationships exist. For example, "A and / or B" can mean: A exists alone, A and B exist simultaneously, or B exists alone. The character " / " generally indicates an "or" relationship between the associated objects.

[0232] The serial numbers of the above embodiments of the present application are for description only and do not represent the advantages or disadvantages of the embodiments.

[0233] The above description is merely an exemplary embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent replacements, improvements, etc. made within the principles of the present application shall be included in the scope of protection of the present application.

Claims

1. An object recommendation method, characterized in that: The method includes: Acquiring reference information, where the reference information includes scene information of multiple scenes, or the reference information includes user information of a target user, at least one of object information of multiple objects, and scene information of the multiple scenes, where the scene represents a type of interactive behavior, where the interactive behavior is an action of a user selecting a resource published by an object in an environment; Determining reference features based on the reference information, where the reference features include scene features of each scene, or the reference features include at least one of a user feature of the target user and an object feature of each object, and the scene features of each scene; Determining an object to be recommended from the multiple objects based on the reference features; Recommending the object to be recommended to the target user; The reference information includes user information of the target user, and determining the reference feature based on the reference information includes: The first feature of the target user is determined based on the user information of the target user, and the first feature of the target user is calculated by the following formula: u=h(e1,e2,…,e F ), where u represents the first feature of the target user, h represents the aggregation function, F represents the amount of attribute information, e1, e2,…, e F The features corresponding to the attribute information are respectively represented, and the features corresponding to the attribute information are calculated using the following formula: ei=1 / (q·xi·Mi), i=1, 2, …, F, where ei represents the feature corresponding to the i-th attribute information, i takes any value from 1 to F, F represents the number of attribute information, q represents the number of non-zero elements in xi, Represents the unique heat encoding or multi-heat encoding of the i-th attribute information, R represents a real number, Represents the dimension of one-hot encoding or multi-hot encoding, represents the embedding matrix of the i-th attribute information, and di×de represents the dimension of the embedding matrix; obtaining at least one first association information, and determining a second feature of the target user based on the at least one first association information, wherein any first association information includes object information of any object selected by the target user, the first association information including determining a target user node, a selected edge, and an object node from a heterogeneous scene hypergraph, and for any target user node, a selected edge, and an object node, using the object information represented by the object node as the first association information; A user feature of the target user is determined based on the first feature of the target user and the second feature of the target user.

2. The method according to claim 1, characterized in that The reference information includes user information of the target user, and determining the reference feature based on the reference information includes: determining a first feature of the target user based on the user information of the target user; Obtaining at least one second association information, and determining a third feature of the target user based on the at least one second association information, wherein any second association information includes scene information of any scene and object information of any object when the target user selects any object in any scene, and the second association information includes determining a target user node, a scene edge, and an object node from a heterogeneous scene hypergraph, and for any target user node, a scene edge, and an object node, determining a piece of second association information using the scene information represented by the scene edge and the object information represented by the object node; A user feature of the target user is determined based on the first feature of the target user and the third feature of the target user.

3. The method according to claim 1, characterized in that The reference information includes user information of the target user, and determining the reference feature based on the reference information includes: determining a first feature of the target user based on the user information of the target user; obtaining at least one third association information, and determining a fourth feature of the target user based on the at least one third association information, wherein any third association information includes scene information of any scene and resource information of any resource when the target user selects any resource in any scene; A user feature of the target user is determined based on the first feature of the target user and the fourth feature of the target user.

4. The method according to claim 1, wherein The reference information includes object information of the plurality of objects, and determining the reference feature based on the reference information includes: For any object, determining a first feature of the object based on the object information of the object; Obtaining at least one fourth association information corresponding to any one of the objects, and determining a second feature of any one of the objects based on the at least one fourth association information corresponding to any one of the objects, wherein any one of the fourth association information corresponding to any one of the objects includes user information of any one of the users when the any one of the objects is selected by the user; An object feature of the any one object is determined based on the first feature of the any one object and the second feature of the any one object.

5. The method according to claim 1, wherein The reference information includes object information of the plurality of objects, and determining the reference feature based on the reference information includes: For any object, determining a first feature of the object based on the object information of the object; obtaining at least one fifth association information corresponding to any one of the objects, and determining a third feature of any one of the objects based on the at least one fifth association information corresponding to any one of the objects, wherein the fifth association information corresponding to any one of the objects includes scene information of any one of the scenes and user information of any one of the users when the object is selected by the user in any one of the scenes; An object feature of the any one object is determined based on the first feature of the any one object and the third feature of the any one object.

6. The method according to claim 1, characterized in that The reference information includes object information of the plurality of objects, and determining the reference feature based on the reference information includes: For any object, determining a first feature of the object based on the object information of the object; Obtaining at least one sixth association information corresponding to any one of the objects, and determining a fourth feature of any one of the objects based on the at least one sixth association information corresponding to any one of the objects, wherein any sixth association information corresponding to any one of the objects includes resource information of any resource published by any one of the objects; An object feature of the any one object is determined based on the first feature of the any one object and the fourth feature of the any one object.

7. The method according to claim 1, characterized in that The reference information includes object information of the plurality of objects, and determining the reference feature based on the reference information includes: For any object, determining a first feature of the object based on the object information of the object; Obtaining at least one seventh association information corresponding to any one of the objects, and determining a fifth feature of any one of the objects based on the at least one seventh association information corresponding to any one of the objects, wherein any seventh association information corresponding to any one of the objects includes, when any one of the objects publishes any one of the resources in any one of the scenarios, scene information of the scene and resource information of the resource; An object feature of the any one object is determined based on the first feature of the any one object and the fifth feature of the any one object.

8. The method according to claim 1, characterized in that The reference information includes scene information of the multiple scenes, and determining the reference feature based on the reference information includes: For any scene, determining a first feature of the scene based on scene information of the scene; Obtaining at least one eighth association information corresponding to any one of the scenarios, and determining a second feature of the any one of the scenarios based on the at least one eighth association information corresponding to the any one of the scenarios, wherein the eighth association information corresponding to the any one of the scenarios includes scenario information of the any one of the scenarios and user information of any one of the users corresponding to the any one of the scenarios; Based on the first feature of the any one scene and the second feature of the any one scene, a scene feature of the any one scene is determined.

9. The method according to claim 1, characterized in that The reference information includes scene information of the multiple scenes, and determining the reference feature based on the reference information includes: For any scene, determining a first feature of the scene based on scene information of the scene; Obtaining at least one ninth association information corresponding to any one of the scenes, and determining a third feature of the any one of the scenes based on the at least one ninth association information corresponding to the any one of the scenes, wherein the ninth association information corresponding to the any one of the scenes includes scene information of the any one of the scenes and object information of any one of the objects corresponding to the any one of the scenes; Based on the first feature of the any one scene and the third feature of the any one scene, a scene feature of the any one scene is determined.

10. The method according to any one of claims 1 to 9, characterized in that: The reference features include user features of the target user, object features of the respective objects, and scene features of the respective scenes; Determining an object to be recommended from the multiple objects based on the reference features includes: Determining a scene to be recommended from the multiple scenes based on the user characteristics of the target user and the scene characteristics of each scene; An object to be recommended is determined from the plurality of objects based on the scene features of the scene to be recommended and the object features of the respective objects.

11. The method according to claim 10, characterized in that The determining of a scene to be recommended from the multiple scenes based on the user characteristics of the target user and the scene characteristics of each scene includes: Acquiring environmental information of a plurality of environments, and determining environmental characteristics of each environment based on the environmental information of each environment; Based on the user characteristics of the target user, the scene characteristics of each scene, and the environmental characteristics of each environment, a scene to be recommended is determined from the multiple scenes.

12. The method according to any one of claims 1 to 9, characterized in that: The determining, based on the reference features, an object to be recommended from the multiple objects includes: Determining index information of each object based on the reference features, where the index information of any object is used to represent the degree of matching between the target user and any object; Based on the indicator information of each object, objects to be recommended whose indicator information meets the screening conditions are screened out from the multiple objects.

13. The method according to any one of claims 1 to 9, characterized in that: The determining of the reference feature based on the reference information includes: Acquire a recommendation model, where the recommendation model includes a feature extraction sub-model and a feature processing sub-model connected in sequence; Determining, by the feature extraction sub-model, a reference feature based on the reference information; The determining, based on the reference features, an object to be recommended from the multiple objects includes: The feature processing sub-model determines an object to be recommended from the multiple objects based on the reference features.

14. An electronic device, characterized in that: The electronic device includes a processor and a memory, wherein at least one program code is stored in the memory, and the at least one program code is loaded and executed by the processor so that the electronic device implements the object recommendation method according to any one of claims 1 to 13.

15. A computer-readable storage medium, characterized in that The computer-readable storage medium stores at least one program code, and the at least one program code is loaded and executed by a processor to enable a computer to implement the object recommendation method according to any one of claims 1 to 13.

16. A computer program product, characterized in that The computer program product stores at least one computer instruction, and the at least one computer instruction is loaded and executed by a processor to enable a computer to implement the object recommendation method according to any one of claims 1 to 13.

Images