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

Abstract

The present disclosure provides an object recommendation method and device, electronic equipment and storage medium, relates to the technical field of artificial intelligence, in particular to the field of big data. The specific implementation scheme is: determining a first similarity between a resource consumption object and a candidate resource production object set according to a first feature vector of the resource consumption object and a second feature vector set of the candidate resource production object set, obtaining at least one first similarity, wherein the first feature vector is used to represent the object attention behavior of the resource consumption object in a plurality of application programs, and the object attention behavior represents the attention behavior of the resource consumption object to the attention resource production object; determining a target resource production object set from the candidate resource production object set according to the at least one first similarity; and recommending the target resource production object set to the resource consumption object.

Description

Technical Field

[0001] This disclosure relates to the field of artificial intelligence technology, and more particularly to the field of big data technology. Specifically, it relates to an object recommendation method, apparatus, electronic device, and storage medium. Background Technology

[0002] With the development of big data technology, massive amounts of relational data have been generated across various industries. Relational data can be used to represent the relationships between entities. Artificial intelligence technology can be used to make object recommendations based on the relational data between entities. Summary of the Invention

[0003] This disclosure provides a method, apparatus, electronic device, and storage medium for object recommendation.

[0004] According to one aspect of this disclosure, an object recommendation method is provided, comprising: determining a first similarity between the resource consumption object and the candidate resource production object set based on a first representation vector of a resource consumption object and a second representation vector set of a candidate resource production object set, thereby obtaining at least one first similarity, wherein the first representation vector is used to represent the object attention behavior of the resource consumption object in multiple applications, and the object attention behavior represents the attention behavior of the resource consumption object towards the resource production object it is interested in; determining a target resource production object set from the candidate resource production object set based on the at least one first similarity; and recommending the target resource production object set to the resource consumption object.

[0005] According to another aspect of this disclosure, an object recommendation apparatus is provided, comprising: a first determining module, configured to determine a first similarity between the resource consumption object and the candidate resource production object set based on a first representation vector of a resource consumption object and a second representation vector set of a candidate resource production object set, thereby obtaining at least one first similarity, wherein the first representation vector is used to represent the object attention behavior of the resource consumption object in multiple applications, and the object attention behavior represents the attention behavior of the resource consumption object towards the resource production object it is interested in; a second determining module, configured to determine a target resource production object set from the candidate resource production object set based on the at least one first similarity; and a recommending module, configured to recommend the target resource production object set to the resource consumption object.

[0006] According to another aspect of this disclosure, an electronic device is provided, comprising: at least one processor; and a memory communicatively connected to said at least one processor; wherein the memory stores instructions executable by said at least one processor, said instructions being executed by said at least one processor to enable said at least one processor to perform the method as described in this disclosure.

[0007] According to another aspect of this disclosure, a non-transitory computer-readable storage medium is provided storing computer instructions, wherein the computer instructions are used to cause the computer to perform the methods described in this disclosure.

[0008] According to another aspect of this disclosure, a computer program product is provided, including a computer program that, when executed by a processor, implements the methods described in this disclosure.

[0009] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of this disclosure, nor is it intended to limit the scope of this disclosure. Other features of this disclosure will become readily apparent from the following description. Attached Figure Description

[0010] The accompanying drawings are provided to better understand this solution and do not constitute a limitation of this disclosure. Wherein:

[0011] Figure 1 This illustration schematically shows an exemplary system architecture for an object recommendation method and apparatus applicable according to embodiments of the present disclosure;

[0012] Figure 2 A flowchart illustrating an object recommendation method according to an embodiment of the present disclosure is shown schematically;

[0013] Figure 3 An example schematic diagram illustrating the determination of a target resource production object set according to an embodiment of the present disclosure is shown.

[0014] Figure 4 An example schematic diagram illustrating the determination of a target resource production object set according to another embodiment of this disclosure is shown;

[0015] Figure 5 An example schematic diagram illustrating the determination of a relationship graph according to an embodiment of the present disclosure is shown.

[0016] Figure 6 An example schematic diagram of a concern relationship diagram according to an embodiment of the present disclosure is shown;

[0017] Figure 7 This schematically illustrates an example diagram of determining a first representation vector according to an embodiment of the present disclosure;

[0018] Figure 8 This schematically illustrates an example diagram of determining a second representation vector according to an embodiment of the present disclosure;

[0019] Figure 9 A block diagram of an object recommendation apparatus according to an embodiment of the present disclosure is schematically shown; and

[0020] Figure 10 A block diagram of an electronic device suitable for implementing an object recommendation method according to an embodiment of the present disclosure is shown schematically. Detailed Implementation

[0021] The exemplary embodiments of this disclosure are described below with reference to the accompanying drawings, including various details of the embodiments to aid understanding, and should be considered merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of this disclosure. Similarly, for clarity and brevity, descriptions of well-known functions and structures are omitted in the following description.

[0022] The collection, storage, use, processing, transmission, provision, and disclosure of user personal information involved in the technical solution disclosed herein comply with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0023] In the technical solution disclosed herein, the user's authorization or consent is obtained before acquiring or collecting the user's personal information.

[0024] Figure 1 The illustration schematically depicts an exemplary system architecture of an object recommendation method and apparatus applicable according to embodiments of the present disclosure.

[0025] It is important to note that Figure 1 The examples shown are merely examples of system architectures applicable to embodiments of this disclosure, intended to help those skilled in the art understand the technical content of this disclosure. They do not imply that embodiments of this disclosure cannot be used in other devices, systems, environments, or scenarios. For instance, in another embodiment, an exemplary system architecture to which the object recommendation processing method and apparatus can be applied may include a terminal device. However, the terminal device can implement the object recommendation method and apparatus provided in the embodiments of this disclosure without interacting with a server.

[0026] like Figure 1 As shown, the system architecture 100 according to this embodiment may include a first terminal device 101, a second terminal device 102, a third terminal device 103, a network 104, and a server 105. The network 104 serves as a medium for providing communication links between the first terminal device 101, the second terminal device 102, the third terminal device 103, and the server 105. The network 104 may include various connection types, such as at least one of wired and wireless communication links. The terminal devices may include at least one of the first terminal device 101, the second terminal device 102, and the third terminal device 103.

[0027] Users can interact with server 105 via network 104 using at least one of the first terminal device 101, second terminal device 102, and third terminal device 103 to receive or send messages, etc. At least one of the first terminal device 101, second terminal device 102, and third terminal device 103 may have various communication client applications installed. For example, at least one of knowledge reading applications, web browser applications, search applications, instant messaging tools, email clients, and social media platform software, etc.

[0028] The first terminal device 101, the second terminal device 102, and the third terminal device 103 can be various electronic devices with displays and supporting web browsing. For example, the electronic devices can include at least one of smartphones, tablets, laptops, and desktop computers.

[0029] Server 105 can be a server that provides various services. For example, Server 105 can be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system. It solves the problems of high management difficulty and weak business scalability in traditional physical hosts and VPS (Virtual Private Server) services.

[0030] It should be noted that the object recommendation method provided in this embodiment of the invention can generally be executed by one of the first terminal device 101, the second terminal device 102, and the third terminal device 103. Correspondingly, the object recommendation device provided in this embodiment of the invention can also be provided in one of the first terminal device 101, the second terminal device 102, and the third terminal device 103.

[0031] Alternatively, the object recommendation method provided in this embodiment of the invention can also generally be executed by server 105. Correspondingly, the object recommendation device provided in this embodiment of the invention can generally be located in server 105. The object recommendation method provided in this embodiment of the invention can also be executed by a server or server cluster that is different from server 105 and capable of communicating with at least one of the first terminal device 101, the second terminal device 102, the third terminal device 103, and server 105. Correspondingly, the object recommendation device provided in this embodiment of the invention can also be located in a server or server cluster that is different from server 105 and capable of communicating with at least one of the first terminal device 101, the second terminal device 102, the third terminal device 103, and server 105.

[0032] It should be understood that Figure 1The number of first terminal devices, second terminal devices, third terminal devices, networks, and servers shown in the diagram is merely illustrative. Depending on implementation needs, any number of first terminal devices, second terminal devices, third terminal devices, networks, and servers can be included.

[0033] It should be noted that the sequence numbers of the operations in the following methods are for descriptive purposes only and should not be considered as indicating the order in which the operations are executed. Unless explicitly stated otherwise, the method does not need to be executed in the exact order shown.

[0034] Figure 2 A flowchart illustrating an object recommendation method according to an embodiment of this disclosure is shown schematically.

[0035] like Figure 2 As shown, the recommended method 200 for this object includes operations S210 to S230.

[0036] In operation S210, based on the first representation vector of the resource consumption object and the second representation vector set of the candidate resource production object set, a first similarity between the resource consumption object and the candidate resource production object set is determined, and at least one first similarity is obtained. The first representation vector is used to represent the object attention behavior of the resource consumption object in multiple applications, and the object attention behavior represents the attention behavior of the resource consumption object to the resource production object it is interested in.

[0037] In operation S220, the target resource production object set is determined from the candidate resource production object set based on at least one first similarity.

[0038] In operation S230, a set of target resource production objects is recommended to the resource consumption objects.

[0039] According to embodiments of this disclosure, objects may include resource consumption objects and resource production objects. Resources may include at least one of the following: articles, videos, and music. Resource consumption objects and resource production objects can be used to represent user identities. Resource consumption objects may refer to users who consume resources. Resource consumption objects can be used to represent users who need to be recommended to objects. Resource production objects may refer to users who produce resources. Resource production objects can be used to represent authors waiting to be recommended to objects. The same user may belong to either a resource consumption object or a resource production object, which can be set according to actual business needs and is not limited here. For example, user X may be a resource consumption object, and user Y may be a resource production object. Alternatively, user X and user Y may be both resource consumption objects and resource production objects.

[0040] According to embodiments of this disclosure, in response to a detected request for object recommendation, a first representation vector of a resource consumption object and a second representation vector set of a candidate resource production object set can be obtained. The candidate resource production object set may include at least one candidate resource production object. The candidate resource production object may be determined from resource production objects. The second representation vector set may include at least one second representation vector. Each of the at least one second representation vector may correspond to a candidate resource production object. The first representation vector can be used to represent the object attention behavior of the resource consumption object in multiple applications. The second representation vector can be used to represent the object attention behavior of the candidate resource production object in multiple applications. Object attention behavior can represent the attention behavior of the resource consumption object towards the resource production object it is interested in. Attention behavior may include at least one of the following: clicking, browsing, favorite, liking, tipping, commenting, bullet comments, sharing, and forwarding.

[0041] According to embodiments of this disclosure, an application can refer to a computer program designed to perform one or more specific tasks. The application can have a visual user interface and be able to interact with the user. Object attention behavior data corresponding to the object attention behavior of a resource consuming object in multiple applications can be obtained. A first representation vector is determined based on the object attention behavior data. The specific method for obtaining the first representation vector can be set according to actual business needs and is not limited here. For example, object attention behavior data corresponding to the object attention behavior of a resource consuming object in multiple applications can be obtained separately using data interfaces corresponding to multiple applications. Alternatively, object attention behavior data corresponding to the object attention behavior of a resource consuming object in multiple applications can be obtained from a database based on the application identifiers corresponding to each of the multiple applications.

[0042] According to embodiments of this disclosure, after obtaining a first representation vector of a resource consumption object and a second representation vector set of a candidate resource production object set, a target resource production object set can be determined from the candidate resource production object set based on the first representation vector of the resource consumption object and the second representation vector set of the candidate resource production object set. The target resource production object set may include at least one target resource production object. A target resource production object may refer to a resource production object that needs to be recommended to a resource consumption object. Target resource production objects can be used to represent similar users corresponding to resource consumption objects. Similar users can be understood as users whose similarity to resource consumption objects meets preset conditions.

[0043] According to embodiments of this disclosure, a first similarity between a resource consumption object and a set of candidate resource production objects can be determined based on a first representation vector and a second set of representation vectors. For example, the vector similarity between each second representation vector in the first and second sets of representation vectors can be determined to obtain at least one first similarity. The first similarity can be used to characterize the degree of similarity between a resource consumption object and a candidate resource production object corresponding to the second representation vector. The specific method for determining the first similarity can be set according to actual business needs and is not limited here. For example, the specific method for determining the first similarity may include at least one of the following: based on cosine similarity, based on Pearson correlation coefficient, based on Euclidean distance, or based on Jaccard distance.

[0044] According to embodiments of this disclosure, after obtaining at least one first similarity, a target resource production object set can be determined from the candidate resource production object set based on the at least one first similarity. For example, at least one candidate resource production object can be sorted based on the at least one first similarity to obtain a sorting result. In this case, the target resource production object set can be determined from the candidate resource production object set based on the sorting result. Alternatively, at least one candidate resource production object can be filtered according to a first preset threshold based on the at least one first similarity to obtain a filtering result. In this case, the target resource production object set can be determined from the candidate resource production object set based on the filtering result. The first preset threshold can be set according to actual business needs and is not limited here. For example, the first preset threshold can be 0.8.

[0045] According to embodiments of this disclosure, since the first representation vector is used to represent the object attention behavior of resource consumption objects in multiple applications, it can strengthen multi-domain attention data, enabling subsequent resource production object recommendations to have interest-point transfer capabilities. Furthermore, since object attention behavior represents the attention behavior of resource consumption objects towards the resource production objects they are interested in, and the second representation vector set is the representation vector set of the candidate resource production object set, it achieves the acquisition of representation vectors along the resource consumption object-resource production object dimension, mitigating the adverse impact of resource dimension bias on resource production object recommendations. Based on this, the target resource production object set is determined from the candidate resource production object set according to the first and second representation vector sets, improving the accuracy of resource production object recommendations.

[0046] This is merely an exemplary embodiment, but it is not limited thereto. Other object recommendation methods known in the art may also be included, as long as they can achieve object recommendation.

[0047] The following is for reference. Figures 3-8 In conjunction with specific embodiments, Figure 2 The method shown will be further explained.

[0048] According to embodiments of this disclosure, the object recommendation method 200 described above may further include the following operations.

[0049] Based on the first representation vector of the resource consumption object and the third representation vector set of the resource consumption object set, a second similarity is determined between the resource consumption object and the resource consumption object set, resulting in at least one second similarity. Based on at least one second similarity, a candidate resource consumption object set is determined from the resource consumption object set. Based on the candidate resource consumption object sets, a candidate resource production object set is determined.

[0050] According to embodiments of this disclosure, in response to a detected request for object recommendation, a first representation vector of a resource-consuming object and a third representation vector set of a set of resource-consuming objects can be obtained. The set of resource-consuming objects may include at least one resource-consuming object. The third representation vector set may include at least one third representation vector. Each of the at least one third representation vector may correspond to a resource-consuming object. The third representation vectors can be used to characterize the object attention behavior of resource-consuming objects across multiple applications.

[0051] According to embodiments of this disclosure, after obtaining a first representation vector of a resource consumption object and a third representation vector set of a resource consumption object set, a second similarity between the resource consumption object and the resource consumption object set can be determined based on the first representation vector and the third representation vector set. After obtaining at least one second similarity, a candidate resource consumption object set can be determined from the resource consumption object set based on the at least one second similarity.

[0052] According to embodiments of this disclosure, at least one resource consumption object can be sorted based on at least one second similarity. A candidate resource consumption object set can be determined from the set of resource consumption objects based on the sorting results. For example, the top 5 second similarities in the sorting results can be selected; in this case, the resource consumption objects corresponding to each of the top 5 second similarities can be determined as the candidate resource consumption object set.

[0053] According to embodiments of this disclosure, at least one resource consumption object can be screened based on at least one second similarity and a second preset threshold. A candidate resource consumption object set can be determined from the set of resource consumption objects based on the screening results. For example, the second preset threshold can be set to 0.6. In this case, resource consumption objects with a second similarity less than 0.6 can be filtered out, and resource consumption objects with a second similarity greater than or equal to 0.6 can be determined as candidate resource consumption objects in the candidate resource consumption object set.

[0054] According to embodiments of this disclosure, since the first representation vector is the representation vector of the resource consumption object and the third representation vector set is the representation vector set of the resource consumption object set, on this basis, by determining the candidate resource consumption object set from the resource consumption object set according to the first representation vector and the third representation vector set, and determining the candidate resource production object set according to the candidate resource consumption object set, the selection of candidate resource production objects is enriched and the understanding of resource consumption objects is improved.

[0055] Figure 3 An example schematic diagram illustrating the determination of a target resource production object set according to an embodiment of the present disclosure is shown.

[0056] like Figure 3 As shown, in response to the detection of an object recommendation instruction, resource consumption object 301 and resource consumption object set 302 can be determined. Based on resource consumption object 301, a first representation vector 303 of resource consumption object 301 is determined. Based on resource consumption object set 302, a third representation vector set 304 of resource consumption object set 302 is determined.

[0057] After determining the first representation vector 303 of resource consumption object 301 and the third representation vector set 304 of resource consumption object set 302, a second similarity between resource consumption object 301 and resource consumption object set 302 can be determined based on the first representation vector 303 of resource consumption object 301 and the third representation vector set 304 of resource consumption object set 302, thus obtaining at least one second similarity. At least one second similarity may include second similarity 305_1, second similarity 305_2, ..., second similarity 305_m, ..., second similarity 305_M. M can be an integer greater than or equal to 1, where m ∈ {1, 2, ..., (M-1), M}.

[0058] After obtaining at least one second similarity, a candidate resource consumption object set 306 can be determined from the resource consumption object set based on the at least one second similarity. Based on the candidate resource consumption object set 306, a candidate resource production object set 307 is determined. Based on the candidate resource production object set 307, a second representation vector set 308 of the candidate resource production object set 307 is determined.

[0059] After determining the first representation vector 303 of the resource consumption object 301 and the second representation vector set 308 of the candidate resource production object set 307, a first similarity between the resource consumption object 301 and the candidate resource production object set 307 can be determined based on the first representation vector 303 of the resource consumption object 301 and the second representation vector set 308 of the candidate resource production object set 307, thus obtaining at least one first similarity. At least one first similarity may include first similarity 305_1, first similarity 305_2, ..., first similarity 305_n, ..., first similarity 305_N. N can be an integer greater than or equal to 1, where n ∈ {1, 2, ..., (N-1), N}.

[0060] After obtaining at least one first similarity, a target resource production object set 310 can be determined from the candidate resource production object set 307 based on the at least one first similarity. After determining the target resource production object set 310, the target resource production object set 310 can be recommended to the resource consumption object 301.

[0061] According to embodiments of this disclosure, the object recommendation method 200 described above may further include the following operations.

[0062] A first set of resource production objects corresponding to resource consumption objects is determined. Based on the fourth representation vector set of the first set of resource production objects and the second representation vector set of the second set of resource production objects, a third similarity is determined between the first set of resource production objects and the second set of resource production objects, resulting in at least one third similarity. Based on at least one third similarity, a candidate set of resource production objects is determined from the second set of resource production objects.

[0063] According to embodiments of this disclosure, in response to detecting a request for object recommendation, a first set of resource production objects corresponding to a resource consumption object can be determined. The first set of resource production objects may include at least one first resource production object. The first resource production object may be an application representing a resource production object directly associated with the current resource consumption object. After determining the first set of resource production objects,

[0064] According to embodiments of this disclosure, after determining a first set of resource production objects, a fourth representation vector set for the first set of resource production objects and a second representation vector set for the second set of resource production objects can be determined. The fourth representation vector set may include at least one fourth representation vector corresponding to each of the first resource production objects. The fourth representation vector can be used to represent the object attention behavior of the first resource production objects in multiple applications. The second representation vector set may include at least one second representation vector corresponding to each of the second resource production objects. For each first resource production object in the first set of resource production objects, each first resource production object can be treated as a resource consumption object, and a second resource production object corresponding to each first resource production object can be determined.

[0065] According to embodiments of this disclosure, after determining the fourth representation vector set of the first resource production object set and the second representation vector set of the second resource production object set, a third similarity between the first resource production object set and the second resource production object set can be determined based on the fourth and second representation vector sets. After obtaining at least one third similarity, a candidate resource production object set can be determined from the second resource production object set based on the at least one third similarity.

[0066] According to embodiments of this disclosure, since the first representation vector is the representation vector of the resource consumption object, the fourth representation vector set is the representation vector set of the resource consumption object set, and the second representation vector set is the representation vector set of the second resource production object set, on this basis, by determining the candidate resource production object set from the second resource production object set according to the fourth and second representation vector sets, the selection of candidate resource production objects is enriched, and the understanding of resource consumption objects is improved.

[0067] Figure 4 An example schematic diagram illustrating the determination of a target resource production object set according to another embodiment of the present disclosure is shown.

[0068] like Figure 4 As shown, in response to the detection of an object recommendation instruction, a resource consumption object 401 can be determined. After determining the resource consumption object 401, a first resource production object set 402 corresponding to the resource consumption object 401 can be determined. After determining the first resource production object set 402, a fourth representation vector set 404 of the first resource production object set 402 can be determined based on the first resource production object set 402. A second representation vector set 405 of the second resource production object set 403 can be determined based on the second resource production object set 403.

[0069] After determining the fourth representation vector set 404 of the first resource production object set 402 and the second representation vector set 405 of the second resource production object set 403, a third similarity between the first resource production object set 402 and the second resource production object set 403 can be determined based on the fourth representation vector set 404 of the first resource production object set 402 and the second representation vector set 405 of the second resource production object set 403, thus obtaining at least one third similarity. At least one third similarity may include third similarity 4061, third similarity 406_2, ..., third similarity 406_p, ..., third similarity 406_P. P can be an integer greater than or equal to 1, where p∈{1, 2, ..., (P-1), P}.

[0070] After obtaining at least one third similarity, the candidate resource production object set 407 can be determined from the second resource production object set 403 based on the at least one third similarity. Based on the candidate resource production object set 407, a second representation vector set 408 of the candidate resource production object set 407 is determined.

[0071] A first representation vector 409 for resource consumption object 401 can be determined. After determining the first representation vector 409 of resource consumption object 401 and the second representation vector set 408 of candidate resource production object set 407, a first similarity between resource consumption object 401 and candidate resource production object set 407 can be determined based on the first representation vector 409 of resource consumption object 401 and the second representation vector set 408 of candidate resource production object set 407, thus obtaining at least one first similarity. At least one first similarity may include first similarity 410_1, first similarity 410_2, ..., first similarity 410_n, ..., first similarity 410_Q. Q can be an integer greater than or equal to 1, where q∈{1, 2, ..., (Q-1), Q}.

[0072] After obtaining at least one first similarity, a target resource production object set 411 can be determined from the candidate resource production object set 407 based on the at least one first similarity. After determining the target resource production object set 411, the target resource production object set 411 can be recommended to the resource consumption object 401.

[0073] According to embodiments of this disclosure, the candidate resource production object set includes at least one candidate resource production object.

[0074] According to embodiments of this disclosure, operation S220 may include the following operations.

[0075] Based on at least one similarity score, at least one candidate resource production object is ranked to obtain first ranking information. Based on the first ranking information, a target resource production object set is determined from the at least one candidate resource production object.

[0076] According to embodiments of this disclosure, after determining at least one similarity, candidate resource production objects corresponding to each of the at least one similarity can be sorted to obtain first sorting information. The first sorting information can be used to characterize the degree of similarity between resource consumption objects and at least one candidate resource production object. A target resource production object set can be determined from the at least one candidate resource production object based on the first sorting information. For example, the top 10 similarities in the first sorting information can be selected; in this case, the candidate resource production objects corresponding to each of the top 10 similarities can be determined as target resource production objects in the target resource production object set.

[0077] According to embodiments of this disclosure, the candidate resource production object set includes at least one candidate resource production object, and the second representation vector set includes second representation vectors corresponding to the at least one candidate resource production object. The first representation vector is obtained by processing the fifth comprehensive object attention behavior data of the resource consumption object node corresponding to the resource consumption object using an attention relationship model. The second representation vector is obtained by processing the sixth comprehensive object attention behavior data of the resource production object node corresponding to the candidate resource production object using an attention relationship model.

[0078] According to embodiments of this disclosure, after determining the resource consumption object and the candidate resource production object set, fifth comprehensive object attention behavior data of the resource consumption object node corresponding to the resource consumption object can be determined based on the resource consumption object. Sixth comprehensive object attention behavior data of the resource production object node corresponding to each candidate resource production object in the candidate resource production object set can be determined based on each candidate resource production object.

[0079] According to embodiments of this disclosure, after determining the fifth comprehensive object attention behavior data of the resource consumption object node corresponding to the resource consumption object, and the sixth comprehensive object attention behavior data of the resource production object node corresponding to each candidate resource production object, the fifth comprehensive object attention behavior data of the resource consumption object node corresponding to the resource consumption object can be processed using an attention relationship model to obtain a first representation vector. The sixth comprehensive object attention behavior data of the resource production object node corresponding to the candidate resource production object can be processed using an attention relationship model to obtain a second representation vector.

[0080] According to embodiments of this disclosure, the attention relationship model is obtained by training a graph neural network model using an attention relationship graph. The attention relationship graph is obtained based on a comprehensive object attention behavior dataset of a resource object set. The attention relationship graph includes multiple resource object nodes and at least one edge. The multiple resource object nodes include at least one resource consuming object node and at least one resource producing object node. The edge represents the attention behavior between the connected resource consuming object node and resource producing object node. The resource object set includes at least one resource object, and the comprehensive object attention behavior dataset includes comprehensive object attention behavior data corresponding to at least one resource object. The comprehensive object attention behavior data is determined based on multiple object attention behavior data, and the object attention behavior data represents the attention behavior data towards resource producing objects when the resource object is a resource consuming object.

[0081] According to embodiments of this disclosure, a resource object set may include at least one resource object. The at least one resource object may include resource consuming objects and resource producing objects. A given resource object may be both a resource consuming object and a resource producing object. Based on each resource object in the resource object set, attention behavior data for resource producing objects can be determined when the resource object is a resource consuming object, to obtain object attention behavior data corresponding to each of the at least one resource object in the resource object set. After determining the object attention behavior data corresponding to each of the at least one resource object, comprehensive object attention behavior data corresponding to each of the at least one resource object can be determined based on the object attention behavior data corresponding to each of the at least one resource object. After determining the comprehensive object attention behavior data corresponding to each of the at least one resource object, a comprehensive object attention behavior dataset can be determined based on the comprehensive object attention behavior data corresponding to each of the at least one resource object.

[0082] According to embodiments of this disclosure, after determining a comprehensive object attention behavior dataset, an attention relationship graph can be determined based on the dataset. For example, at least one resource object in the resource object set can be designated as a resource object node. When attention behavior exists between two resource objects, the resource object nodes corresponding to those two resource objects can be connected by edges, thereby generating an attention relationship graph. The attention relationship graph may include multiple resource object nodes and at least one edge. The multiple resource object nodes may include at least one resource consuming object node and at least one resource producing object node. Edges can be used to represent the attention behavior between the connected resource consuming object node and resource producing object node.

[0083] According to embodiments of this disclosure, the attention relationship model can be obtained by training a graph neural network (GNN) model using an attention relationship graph. The model structure of the graph neural network model can be configured according to actual business needs and is not limited herein. The graph neural network model can include at least one of the following: graph convolutional network (GCN), graph auto-encoder (GAE), graph generative network (GGN), graph recurrent network (GRN), and graph attention network (GAT). The training method of the graph neural network model can be configured according to actual business needs and is not limited herein. For example, the training method can include at least one of the following: unsupervised training, supervised training, and semi-supervised training.

[0084] According to embodiments of this disclosure, taking a graph neural network model as an example of a graph convolutional network, the graph convolutional network may include GraphSAGE (i.e., Graph Sample and Aggregate). GraphSAGE may include a feature function (i.e., Sample) and an aggregation function (i.e., Aggregate). The feature function can be used to sample other resource object nodes associated with the current resource object node according to preset rules to determine the feature vectors of other resource object nodes associated with the current resource object node. The feature function may employ a fixed-length sampling method, defining the required number of other resource object nodes associated with the current resource object node, and using a resampling with replacement or a negative sampling method to obtain the feature vectors of other resource object nodes associated with the current resource object node. Based on this, the aggregation function can be used to aggregate the feature vectors of other resource object nodes associated with the current resource object node to obtain an updated aggregated vector of the current resource object node.

[0085] According to embodiments of this disclosure, the feature vector of the current resource object node from the previous iteration and the updated representation vector of the current resource object node in the current iteration can be fused to obtain a fusion result. Based on this, the fusion result can be input to a fully connected layer, outputting the representation vector of the current resource object node. The specific structure of the fully connected layer can be configured according to actual business needs and is not limited here. For example, a fully connected layer based on a nonlinear activation function can be used. The nonlinear activation function can include at least one of the following: the Sigmoid function, the hyperbolic tangent function (Tanh), the modified linear unit function (ReLU), the leaky rectified linear unit function (LReLU), the exponential linear unit function (ELU), the Gaussian error linear unit function (GeLu), and the Softmax function.

[0086] According to embodiments of this disclosure, an attention relationship model is obtained by training a graph neural network model using an attention relationship graph, thereby improving the generalization ability of the attention relationship model and further expanding the applicable scenarios of the object recommendation method. Based on this, a first representation vector is obtained by processing the fifth comprehensive object attention behavior data of the resource consumption object node corresponding to the resource consumption object using the attention relationship model, and a second representation vector is obtained by processing the sixth comprehensive object attention behavior data of the resource production object node corresponding to the candidate resource production object using the attention relationship model. This improves the representational power of the first and second representation vectors and enhances the recommendation accuracy of resource production objects.

[0087] According to embodiments of this disclosure, the resource object set is determined from a plurality of candidate resource objects based on a sample selection strategy and a candidate object attention behavior dataset. The sample selection strategy includes at least one of a time-based strategy, a resource consumption object-based strategy, and a resource production object-based strategy. The candidate object attention behavior dataset includes multiple candidate object attention behavior data, each including the time of the candidate object attention behavior. The time-based strategy is used to determine whether the time of the candidate object attention behavior meets predetermined conditions. The resource consumption object-based strategy is used to evaluate the activity level of attention behavior when the candidate resource object is a resource consumption object. The resource production object-based strategy is used to evaluate the object quality level when the candidate resource object is a resource production object.

[0088] According to embodiments of this disclosure, a sliding time window of a preset duration can be constructed according to a time series. Whenever the sliding time window moves one step in the same direction along the path, resource consumption object nodes within the sliding time window and corresponding global attention behavior data can be acquired. Global attention behavior data can refer to object attention behavior data corresponding to multiple applications. The direction of movement can refer to the direction away from the starting position. After each step the sliding time window moves, the central resource consumption object node at the center of the sliding time window can be combined with the other resource consumption object nodes within the sliding time window to obtain training samples.

[0089] According to embodiments of this disclosure, the candidate object attention behavior dataset may include multiple candidate object attention behavior data. The candidate object attention behavior data may include the time of the candidate object attention behavior. The sample selection strategy may include at least one of the following: a time-dimensional strategy, a resource-consuming object-dimensional strategy, and a resource-producing object-dimensional strategy. The time-dimensional strategy can be used to determine whether the time of the candidate object attention behavior meets predetermined conditions. The time-dimensional strategy and preset conditions can be set according to actual business needs and are not limited herein. For example, the preset condition may be that the interval from the current system time is greater than 12 hours. In this case, the time-dimensional strategy may include determining the difference between the current system time and the time of the candidate object attention behavior; if the difference is greater than 12 hours, it can be determined that the time of the candidate object attention behavior meets the predetermined conditions.

[0090] According to embodiments of this disclosure, when a candidate resource object is a resource consumption object, the activity level of attention behavior can be assessed based on a resource consumption object-dimensional strategy. The resource consumption object-dimensional strategy can be set according to actual business needs and is not limited thereto. For example, the resource consumption object-dimensional strategy may include determining the frequency of attention behavior for a resource consumption object, and determining that the activity level of the resource consumption object's attention behavior is at a higher level when the frequency of attention behavior reaches a third preset threshold.

[0091] According to embodiments of this disclosure, when a candidate resource object is a resource production object, since the resource production object can continuously produce resources, and the resources produced by the resource production object each time may not have obvious correlation or consistency, the quality of the object can be evaluated based on a resource production object-level strategy to address the high debiasing cost at the resource production object level.

[0092] According to embodiments of this disclosure, a resource production object dimension strategy can achieve debiasing of the resource production object dimension. The resource production object dimension strategy can be set according to actual business needs and is not limited herein. The resource production object dimension strategy may also include at least one of the following: determining the resource production activity level of a resource production object, determining the verticality of the production resources of a resource production object, determining the number of resource consumption objects corresponding to the resource production object, and determining the number of newly added resource consumption objects corresponding to the resource production object. For example, the resource production object dimension strategy may include determining the number of resource consumption objects corresponding to a resource production object, and if the number of resource consumption objects corresponding to a resource production object reaches a fourth preset threshold, determining that the object quality level of the resource production object is a higher level.

[0093] According to embodiments of this disclosure, a resource object set is determined based on a sample selection strategy and a candidate object attention behavior dataset, thereby enabling the selection of training samples for the attention relationship model. Furthermore, since the sample selection strategy includes time-dimensional, resource-consuming object-dimensional, and resource-producing object-dimensional strategies, the exposure rate of fresh resources is increased, enhancing the effectiveness of the attention relationship model.

[0094] According to embodiments of this disclosure, multiple object attention behavior data are obtained using data interfaces corresponding to multiple applications.

[0095] According to embodiments of this disclosure, object attention behavior data corresponding to each of multiple applications can be obtained using data interfaces corresponding to each application. For example, object attention behavior data corresponding to different applications can be set according to a predetermined format strategy. The predetermined format strategy may include one of the following: object attention behavior data format, object attention behavior data meaning, and object attention behavior data storage location.

[0096] According to embodiments of this disclosure, the meaning of object attention behavior data can refer to the need to ensure that the meaning of object attention behavior data is consistent across different applications for the same type of attention behavior. For example, application A generates an attention behavior through a long press, while application B generates an attention behavior through clicking on an avatar. In this case, it is necessary to ensure that the data representing clicking on an avatar in application A and the data representing a long press in application B have consistent meanings.

[0097] Figure 5 An example schematic diagram illustrating the determination of a concern relationship graph according to an embodiment of the present disclosure is shown.

[0098] like Figure 5As shown, a candidate object attention behavior dataset 502 can be determined based on the candidate resource object set 501. Based on the sample selection strategy 503 and the candidate object attention behavior dataset 502 of the candidate resource object set 501, a resource object set 504 is determined from the multiple candidate resource objects included in the candidate resource object set 501.

[0099] The resource object set 504 may include at least one resource object. At least one resource object may include resource object 504_1, resource object 504_2, ..., resource object 504_x, ..., resource object 504_X. X may be an integer greater than or equal to 1, x∈{1, 2, ..., (X-1), X}.

[0100] After determining resource object set 504, if resource object 504_1 is a resource consumer object, the attention behavior data 505_1 for resource object 504_1 can be determined. If resource object 504_2 is a resource consumer object, the attention behavior data 505_2 for resource object 504_2 can be determined. If resource object 504_x is a resource consumer object, the attention behavior data 505_x for resource object 504_x can be determined. If resource object 504X is a resource consumer object, the attention behavior data 505_X for resource object 504_X can be determined.

[0101] Based on the attention behavior data 505_1, attention behavior data 505_2, ..., attention behavior data 505_x, ..., attention behavior data 505_X, a comprehensive object attention behavior dataset 506 can be determined. After determining the comprehensive object attention behavior dataset 506, an attention relationship graph 507 can be obtained based on the comprehensive object attention behavior dataset 506 of the resource object set 504.

[0102] Figure 6 An example schematic diagram of a concern relationship diagram according to an embodiment of the present disclosure is shown.

[0103] like Figure 6As shown, the attention graph 600 may include multiple resource object nodes and at least one edge. The multiple resource object nodes may include at least resource object nodes n_601, n_602_1, n_602_2, n_602_3, n_602_4, n_602_5, n_603_1, n_603_2, n_603_3, n_603_4, and n_603_5. The at least one edge may, for example, include edges e_601_1, e_602_1, e_602_2, and e_602_3. Edges between any other two resource object nodes in the attention graph 600 will not be described further here.

[0104] Focusing on relational graph 600, taking resource object node n_601 as a resource consumer object node as an example, in this case, resource object nodes n_602_1, n_602_2, n_602_3, n_602_4, and n_602_5 can serve as first-order resource producer object nodes of resource object node n_601. Resource object nodes n_603_1, n_603_2, n_603_3, n_603_4, and n_603_5 can serve as second-order resource producer object nodes of resource object node n_601.

[0105] For resource object nodes n_601, n_602_1, and n_603_5, edge e_601_1 can be used to represent the attention behavior between resource object nodes n_601 and n_602_1, and edge e_602_1 can be used to represent the attention behavior between resource object nodes n_602_1 and n_603_5. In this case, resource object node n_602_1 itself can act as both a resource consumer node and a resource producer node of resource object node n_601.

[0106] For resource object nodes n_602_3, n_602_4, and n_603_4, edge e_602_2 can be used to represent the attention behavior between resource object nodes n_602_3 and n_603_4, and edge e_602_3 can be used to represent the attention behavior between resource object nodes n_602_4 and n_603_4. In this case, resource object node n_603_4 can act as both a resource production object node for resource object node n_602_3 and a resource production object node for resource object node n_602_4.

[0107] According to embodiments of this disclosure, the first representation vector is determined based on first integrated object attention behavior data and second integrated object attention behavior data of first neighbor resource production objects. The first integrated object attention behavior data is determined based on multiple first object attention behavior data sets, each corresponding to a multiple application. The first neighbor resource production object is determined based on the first integrated object attention behavior data. The second integrated object attention behavior data is determined based on multiple second object attention behavior data sets, each corresponding to a multiple application.

[0108] According to embodiments of this disclosure, a first neighboring resource production object corresponding to a resource consumption object can be directly determined based on first comprehensive object attention behavior data. For example, first comprehensive object attention behavior data can be determined based on multiple first object attention behavior data. Each of the multiple first object attention behavior data can be used to characterize the object attention behavior data of a resource consumption object corresponding to an application. A first neighboring resource production object corresponding to a resource consumption object can be determined based on the first comprehensive object attention behavior data. In this case, the first neighboring resource production object may include at least one first-order neighboring resource production object.

[0109] According to embodiments of this disclosure, alternatively, the first neighboring resource-producing object corresponding to the resource-consuming object can also be determined indirectly based on the first comprehensive object attention behavior data. In this case, the first neighboring resource-producing object may include first-order neighboring resource-producing objects and second-order neighboring resource-producing objects.

[0110] According to embodiments of this disclosure, second comprehensive object attention behavior data can be determined based on a plurality of second object attention behavior data. Each of the plurality of second object attention behavior data can be used to characterize the object attention behavior data of a first neighboring resource-producing object corresponding to the application. A first representation vector of a resource-consuming object can be determined based on the first comprehensive object attention behavior data and the second comprehensive object attention behavior data of the first neighboring resource-producing objects.

[0111] According to embodiments of this disclosure, the first representation vector is determined based on first fused feature data. The first fused feature data is obtained by fusing first aggregated feature data and behavioral feature data corresponding to the first integrated object attention behavior data. The first fused feature data is also obtained by feature aggregation of second integrated object attention behavior data.

[0112] According to embodiments of this disclosure, first aggregated feature data is obtained by aggregating features of the second comprehensive object attention behavior data of the first neighboring resource production object. First fused feature data is obtained by fusing the first aggregated feature data and the behavior feature data corresponding to the first comprehensive object attention behavior data. Based on this, a first representation vector of the resource consumption object can be obtained from the first fused feature data.

[0113] According to embodiments of this disclosure, taking GraphSAGE as an example, GraphSAGE includes feature functions and aggregation functions. Feature functions can be used to extract features from first neighboring resource-producing objects to determine second comprehensive object attention behavior data of the first neighboring resource-producing objects. Aggregation functions can be used to aggregate features from the second comprehensive object attention behavior data of the first neighboring resource-producing objects to obtain first aggregated feature data. The first aggregated feature data and the behavioral feature data corresponding to the first comprehensive object attention behavior data can be fused to obtain first fused feature data. Based on this, the first fused feature data can be input into a fully connected layer to output a first representation vector of the resource-consuming object.

[0114] According to embodiments of this disclosure, the first comprehensive object attention behavior data is obtained by processing multiple first object attention behavior data using a predetermined format strategy, and the multiple first object attention behavior data have the same data format.

[0115] According to embodiments of this disclosure, multiple first object attention behavior data can be processed using a predetermined format strategy to obtain first comprehensive object attention behavior data. The predetermined format strategy can be set according to actual business needs and is not limited thereto. For example, the predetermined format strategy may include one of the following: object attention behavior data format, object attention behavior data meaning, and object attention behavior data storage location.

[0116] According to embodiments of this disclosure, since the first integrated object attention behavior data is obtained by processing multiple first object attention behavior data using a predetermined format strategy, a first representation vector is determined based on the first integrated object attention behavior data and the second integrated object attention behavior data of the first neighboring resource production object, thereby improving the understanding of resource consumption objects.

[0117] Figure 7 The illustration shows an example schematic diagram of determining a first representation vector according to an embodiment of the present disclosure.

[0118] like Figure 7 As shown, first comprehensive object attention behavior data 702 can be determined based on multiple first object attention behavior data 701. Based on the first comprehensive object attention behavior data 702, behavioral feature data 703 corresponding to the first comprehensive object attention behavior data 702 can be determined.

[0119] Based on multiple second object attention behavior data 704, second comprehensive object attention behavior data 705 can be determined. Based on the second comprehensive object attention behavior data 705, first aggregated feature data 706 corresponding to the first comprehensive object attention behavior data 705 can be determined.

[0120] After determining the behavioral feature data 703 and the first aggregated feature data 706 corresponding to the first comprehensive object attention behavior data 702, they can be fused to obtain the first fused feature data 707. The first representation vector 708 can then be determined based on the first fused feature data 707.

[0121] According to embodiments of this disclosure, the candidate resource production object set includes at least one candidate resource production object, and the second representation vector set includes a second representation vector corresponding to the at least one candidate resource production object. The second representation vector is determined based on third integrated object attention behavior data and fourth integrated object attention behavior data of second neighboring resource production objects. The third integrated object attention behavior data is determined based on multiple third object attention behavior data, which correspond to multiple applications. The second neighboring resource production objects are determined based on the third integrated object attention behavior data. The fourth integrated object attention behavior data is determined based on multiple fourth object attention behavior data, which correspond to multiple applications.

[0122] According to embodiments of this disclosure, a second neighboring resource production object corresponding to a resource consumption object can be directly determined based on third comprehensive object attention behavior data. For example, for a candidate resource production object among at least one candidate resource production object, third comprehensive object attention behavior data can be determined based on multiple third object attention behavior data. Based on the third comprehensive object attention behavior data of the candidate resource production object, a second neighboring resource production object corresponding to the candidate resource production object is determined. In this case, the second neighboring resource production object may include at least one first-order candidate neighboring resource production object.

[0123] According to embodiments of this disclosure, alternatively, a second neighboring resource producing object corresponding to a candidate resource consuming object can also be determined indirectly based on third-level comprehensive object attention behavior data. In this case, the second neighboring resource producing object may include first-order candidate neighboring resource producing objects and second-order candidate neighboring resource producing objects.

[0124] According to embodiments of this disclosure, fourth comprehensive object attention behavior data can be determined based on a plurality of fourth object attention behavior data. Each of the plurality of fourth object attention behavior data can be used to characterize the object attention behavior data of a second neighbor resource production object corresponding to the application. A second representation vector of a candidate resource consumption object can be determined based on the second comprehensive object attention behavior data and the fourth comprehensive object attention behavior data of the second neighbor resource production object.

[0125] According to embodiments of this disclosure, the second representation vector is obtained based on second fused feature data. The second fused feature data is obtained by fusing second aggregated feature data and behavioral feature data corresponding to the third integrated object attention behavior data. The second aggregated feature data is obtained by feature aggregation of the fourth integrated object attention behavior data.

[0126] According to embodiments of this disclosure, second aggregated feature data is obtained by aggregating features of the fourth comprehensive object attention behavior data of the second neighbor resource production object. Second fused feature data is obtained by fusing the second aggregated feature data with the behavior feature data corresponding to the third comprehensive object attention behavior data. A second representation vector of the candidate resource production object is obtained based on the second fused feature data.

[0127] According to embodiments of this disclosure, taking GraphSAGE as an example, GraphSAGE includes feature functions and aggregation functions. Feature functions can be used to extract features from second-neighbor resource-producing objects to determine the fourth comprehensive object attention behavior data of the second-neighbor resource-producing objects. Aggregation functions can be used to aggregate features from the fourth comprehensive object attention behavior data of the second-neighbor resource-producing objects to obtain second aggregated feature data. The second aggregated feature data and the behavioral feature data corresponding to the third comprehensive object attention behavior data can be fused to obtain second fused feature data. Based on this, the second fused feature data can be input into a fully connected layer to output a second representation vector of the candidate resource-producing objects.

[0128] According to embodiments of this disclosure, the third comprehensive object attention behavior data is obtained by processing multiple third object attention behavior data using a predetermined format strategy, and the multiple third object attention behavior data have the same data format.

[0129] According to embodiments of this disclosure, multiple third-party attention behavior data can be processed using a predetermined format strategy to obtain third-comprehensive object attention behavior data. The predetermined format strategy can be set according to actual business needs and is not limited herein. For example, the predetermined format strategy may include one of the following: object attention behavior data format, object attention behavior data meaning, and object attention behavior data storage location.

[0130] According to embodiments of this disclosure, since the third integrated object attention behavior data is obtained by processing multiple third object attention behavior data using a predetermined format strategy, the second representation vector is determined based on the third integrated object attention behavior data and the fourth integrated object attention behavior data of the second neighbor resource production object, thereby improving the understanding of the resource production object.

[0131] Figure 8 An example schematic diagram illustrating the determination of a second representation vector according to an embodiment of the present disclosure is shown.

[0132] like Figure 8 As shown, third comprehensive object attention behavior data 802 can be determined based on multiple third object attention behavior data 801. Based on the third comprehensive object attention behavior data 802, behavioral feature data 803 corresponding to the third comprehensive object attention behavior data 802 can be determined.

[0133] Based on multiple fourth object attention behavior data 804, fourth comprehensive object attention behavior data 805 can be determined. Based on fourth comprehensive object attention behavior data 805, second aggregated feature data 806 corresponding to fourth comprehensive object attention behavior data 805 can be determined.

[0134] After determining the behavioral feature data 803 and the second aggregated feature data 806 corresponding to the fourth integrated object's attention behavior data 802, they can be fused to obtain the second fused feature data 807. The second representation vector 808 can then be determined based on the second fused feature data 807.

[0135] According to embodiments of this disclosure, the object recommendation method 200 described above may further include the following operations.

[0136] Identification information matching the first identification information of the resource consumption object is determined from the set of association relationships. The representation vector corresponding to the identification information matching the first identification information is determined as the first representation vector.

[0137] According to embodiments of this disclosure, the object recommendation method 200 described above may further include the following operations.

[0138] The set of associations includes at least one association, and the association represents the relationship between the identifier information and the representation vector.

[0139] According to embodiments of this disclosure, identification information can be used to identify the representation vectors of different resource consumption objects. The specific form of the identification information can be set according to actual business needs and is not limited herein. For example, the identification information may include a unique code (Identity Document, ID) corresponding to the resource consumption object in the application.

[0140] According to embodiments of this disclosure, the attention behavior data corresponding to each of the multiple resource consumption objects can be processed in advance using an attention relationship model to obtain a representation vector corresponding to each of the multiple resource consumption objects. The association relationships corresponding to each of the multiple resource consumption objects can be determined based on their unique codes and representation vectors. On this basis, a set of association relationships can be determined based on the association relationships corresponding to each of the multiple resource consumption objects.

[0141] According to embodiments of this disclosure, after determining the association set, the association set can be stored. The specific storage method can be set according to actual business needs and is not limited here. For example, the association set can be stored in external storage so that, during the recall and sorting stages, the first representation vector corresponding to the resource consumption object can be retrieved from the external storage based on the identification information. Alternatively, the association set can be stored in a server so that, during the recall and sorting stages, the first representation vector corresponding to the resource consumption object can be retrieved from the server based on the identification information using multi-path recall.

[0142] According to embodiments of this disclosure, the object recommendation method 200 described above may further include the following operations.

[0143] From the set of association relationships, determine the set of identifiers that matches the second set of identifiers for the candidate resource production object set. Then, determine the set of representation vectors corresponding to the set of identifiers that matches the second set of identifiers as the second set of representation vectors.

[0144] According to embodiments of this disclosure, the second identification information set may include at least one second identification information. Each of the at least one second identification information can be used to identify a candidate resource production object. The association set may be stored in external memory to facilitate the retrieval of second representation vectors corresponding to each of the multiple candidate resource production objects from the external memory based on multiple-way recall during the recall and ranking stages. A second representation vector set is determined based on the second representation vectors corresponding to each of the multiple candidate resource production objects. Alternatively, the association set may be stored in a server to facilitate the retrieval of second representation vectors corresponding to each of the multiple candidate resource production objects from the server based on multiple-way recall during the recall and ranking stages. A second representation vector set is determined based on the second representation vectors corresponding to each of the multiple candidate resource production objects.

[0145] According to embodiments of this disclosure, since the first representation vector is determined by identifying the identification information that matches the first identification information of the resource consumption object from the association set, and the second representation vector is determined by identifying the identification information set that matches the second identification information set of the candidate resource production object set from the association set, the efficiency and accuracy of object recommendation are improved.

[0146] Figure 7 A block diagram of an object recommendation apparatus according to an embodiment of the present disclosure is shown schematically.

[0147] like Figure 7 As shown, the object recommendation device 700 includes a first determination module 710, a second determination module 720, and a recommendation module 730.

[0148] The first determining module 710 is used to determine a first similarity between a resource consumption object and a candidate resource production object set based on a first representation vector of the resource consumption object and a second representation vector set of the candidate resource production object set, thereby obtaining at least one first similarity. The first representation vector is used to represent the object attention behavior of the resource consumption object in multiple applications, and the object attention behavior represents the attention behavior of the resource consumption object to the resource production object it is interested in.

[0149] The second determining module 720 is used to determine the target resource production object set from the candidate resource production object set based on at least one first similarity.

[0150] Recommendation module 730 is used to recommend a set of target resource production objects to resource consumers.

[0151] According to embodiments of this disclosure, the object recommendation device 700 may further include a third determination module, a fourth determination module, and a fifth determination module.

[0152] The third determining module is used to determine the second similarity between the resource consumption object and the resource consumption object set based on the first representation vector of the resource consumption object and the third representation vector set of the resource consumption object set, thereby obtaining at least one second similarity.

[0153] The fourth determining module is used to determine a candidate set of resource consumption objects from the set of resource consumption objects based on at least one second similarity.

[0154] The fifth determination module is used to determine the candidate resource production object set based on the candidate resource consumption object set.

[0155] According to embodiments of this disclosure, the object recommendation device 700 may further include a sixth determination module, a seventh determination module, and an eighth determination module.

[0156] The sixth determination module is used to determine the first set of resource production objects corresponding to the resource consumption objects.

[0157] The seventh determining module is used to determine the third similarity between the first resource production object set and the second resource production object set based on the fourth representation vector set of the first resource production object set and the second representation vector set of the second resource production object set, thereby obtaining at least one third similarity.

[0158] The eighth determining module is used to determine a candidate resource production object set from the second resource production object set based on at least one third similarity.

[0159] According to embodiments of this disclosure, the candidate resource production object set includes at least one candidate resource production object.

[0160] According to embodiments of this disclosure, the second determining module 720 may include a sorting unit and a determining unit.

[0161] A sorting unit is used to sort at least one candidate resource production object according to at least one similarity to obtain first sorting information.

[0162] The determining unit is used to determine a set of target resource production objects from at least one candidate resource production object based on the first sorting information.

[0163] According to embodiments of this disclosure, the object recommendation device 700 may further include a ninth determination module and a tenth determination module.

[0164] The ninth determination module is used to determine the identification information that matches the first identification information of the resource consumption object from the set of association relationships.

[0165] The tenth determining module is used to determine the representation vector corresponding to the identification information that matches the first identification information as the first representation vector.

[0166] According to embodiments of this disclosure, the association set includes at least one association, and the association represents the representation vector corresponding to the relationship between the identification information and the representation vector.

[0167] According to embodiments of this disclosure, the object recommendation device 700 may further include an eleventh determination module and a twelfth determination module.

[0168] The eleventh determination module is used to determine the set of identification information that matches the second set of identification information of the candidate resource production object set from the set of association relationships.

[0169] The twelfth determining module is used to determine the representation vector set corresponding to the identification information set that matches the second identification information set as the second representation vector set.

[0170] According to embodiments of this disclosure, the first representation vector is determined based on first integrated object attention behavior data and second integrated object attention behavior data of the first neighbor resource production object.

[0171] According to embodiments of this disclosure, the first comprehensive object attention behavior data is determined based on a plurality of first object attention behavior data, which correspond to a plurality of applications.

[0172] According to embodiments of this disclosure, the first neighbor resource production object is determined based on the first comprehensive object attention behavior data.

[0173] According to embodiments of this disclosure, the second comprehensive object attention behavior data is determined based on a plurality of second object attention behavior data, which correspond to a plurality of applications.

[0174] According to embodiments of this disclosure, the first representation vector is determined based on first fused feature data.

[0175] According to embodiments of this disclosure, the first fused feature data is obtained by fusing the first aggregated feature data and the behavioral feature data corresponding to the first comprehensive object attention behavior data.

[0176] According to embodiments of this disclosure, the first fused feature data is obtained by aggregating features from the second comprehensive object attention behavior data.

[0177] According to embodiments of this disclosure, the first comprehensive object attention behavior data is obtained by processing multiple first object attention behavior data using a predetermined format strategy, and the multiple first object attention behavior data have the same data format.

[0178] According to embodiments of this disclosure, the candidate resource production object set includes at least one candidate resource production object, and the second representation vector set includes a second representation vector corresponding to the at least one candidate resource production object.

[0179] According to embodiments of this disclosure, the second representation vector is determined based on third integrated object attention behavior data and fourth integrated object attention behavior data of second neighbor resource production objects.

[0180] According to embodiments of this disclosure, the third comprehensive object attention behavior data is determined based on multiple third object attention behavior data, which correspond to multiple applications.

[0181] According to embodiments of this disclosure, the second neighbor resource production object is determined based on the attention behavior data of the third integrated object.

[0182] According to embodiments of this disclosure, the fourth comprehensive object attention behavior data is determined based on a plurality of fourth object attention behavior data, which correspond to a plurality of applications.

[0183] According to embodiments of this disclosure, the second representation vector is obtained based on the second fused feature data.

[0184] According to embodiments of this disclosure, the second fused feature data is obtained by fusing the second aggregated feature data and the behavioral feature data corresponding to the third comprehensive object attention behavior data.

[0185] According to embodiments of this disclosure, the second aggregated feature data is obtained by aggregating features from the fourth comprehensive object attention behavior data.

[0186] According to embodiments of this disclosure, the third comprehensive object attention behavior data is obtained by processing multiple third object attention behavior data using a predetermined format strategy, and the multiple third object attention behavior data have the same data format.

[0187] According to embodiments of this disclosure, the candidate resource production object set includes at least one candidate resource production object, and the second representation vector set includes a second representation vector corresponding to the at least one candidate resource production object.

[0188] According to an embodiment of this disclosure, the first representation vector is obtained by processing the fifth comprehensive object attention behavior data of the resource consumption object node corresponding to the resource consumption object using an attention relationship model.

[0189] According to an embodiment of this disclosure, the second representation vector is obtained by processing the sixth comprehensive object attention behavior data of the resource production object node corresponding to the candidate resource production object using an attention relationship model.

[0190] According to embodiments of this disclosure, the attention relationship model is obtained by training a graph neural network model using an attention relationship graph.

[0191] According to embodiments of this disclosure, the attention graph is obtained from a comprehensive object attention behavior dataset of a set of resource objects. The attention graph includes multiple resource object nodes and at least one edge. The multiple resource object nodes include at least one resource consuming object node and at least one resource producing object node. The edge is used to represent the attention behavior between the connected resource consuming object node and resource producing object node.

[0192] According to embodiments of this disclosure, the resource object set includes at least one resource object, and the comprehensive object attention behavior dataset includes comprehensive object attention behavior data corresponding to at least one resource object. The comprehensive object attention behavior data is determined based on multiple object attention behavior data, and the object attention behavior data represents the attention behavior data of the resource production object when the resource object is a resource consumption object.

[0193] According to embodiments of this disclosure, the resource object set is determined from multiple candidate resource objects included in the candidate resource object set based on a sample screening strategy and a candidate object attention behavior dataset of the candidate resource object set.

[0194] According to embodiments of this disclosure, the sample selection strategy includes at least one of a time-based strategy, a resource consumption object-based strategy, and a resource production object-based strategy.

[0195] According to embodiments of this disclosure, the candidate object attention behavior dataset includes multiple candidate object attention behavior data, and the candidate object attention behavior data includes candidate object attention behavior time. A time dimension strategy is used to determine whether the candidate object attention behavior time meets a predetermined condition. A resource consumption object dimension strategy is used to evaluate the activity level of attention behavior when the candidate resource object is a resource consumption object. A resource production object dimension strategy is used to evaluate the object quality level when the candidate resource object is a resource production object.

[0196] According to embodiments of this disclosure, multiple object attention behavior data are obtained using data interfaces corresponding to multiple applications.

[0197] According to embodiments of this disclosure, this disclosure also provides an electronic device, a readable storage medium, and a computer program product.

[0198] According to an embodiment of the present disclosure, an electronic device includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the method as described in the present disclosure.

[0199] According to embodiments of the present disclosure, a non-transitory computer-readable storage medium stores computer instructions, wherein the computer instructions are used to cause a computer to perform the methods described in the present disclosure.

[0200] According to embodiments of the present disclosure, a computer program product includes a computer program that, when executed by a processor, implements the methods described in the present disclosure.

[0201] Figure 10A block diagram schematically illustrates an electronic device suitable for implementing an object recommendation method according to embodiments of the present disclosure. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0202] like Figure 10 As shown, the electronic device 1000 includes a computing unit 1001, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 1002 or a computer program loaded from a storage unit 1008 into a random access memory (RAM) 1003. The RAM 1003 may also store various programs and data required for the operation of the electronic device 1000. The computing unit 1001, ROM 1002, and RAM 1003 are interconnected via a bus 1004. An input / output (I / O) interface 1005 is also connected to the bus 1004.

[0203] Multiple components in electronic device 1000 are connected to I / O interface 1005, including: input unit 1006, such as keyboard, mouse, etc.; output unit 1007, such as various types of displays, speakers, etc.; storage unit 1008, such as disk, optical disk, etc.; and communication unit 1009, such as network card, modem, wireless transceiver, etc. Communication unit 1009 allows electronic device 1000 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0204] The computing unit 1001 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 1001 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 1001 performs the various methods and processes described above, such as the object recommendation method. For example, in some embodiments, the object recommendation method may be implemented as a computer software program tangibly contained in a machine-readable medium, such as storage unit 1008. In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device 1000 via ROM 1002 and / or communication unit 1009. When the computer program is loaded into RAM 1003 and executed by the computing unit 1001, one or more steps of the object recommendation method described above may be performed. Alternatively, in other embodiments, the computing unit 1001 may be configured to perform the object recommendation method by any other suitable means (e.g., by means of firmware).

[0205] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0206] The program code used to implement the methods of this disclosure may be written in any combination of one or more programming languages. This program code may be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing apparatus, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code may be executed entirely on a machine, partially on a machine, as a standalone software package partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0207] In the context of this disclosure, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0208] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor); and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the computer. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0209] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as a data server), or computing systems that include middleware components (e.g., an application server), or computing systems that include frontend components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with embodiments of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., a communication network). Examples of communication networks include local area networks (LANs), wide area networks (WANs), and the Internet.

[0210] Computer systems can include clients and servers. Clients and servers are generally located far apart and typically interact via communication networks. Client-server relationships are created by computer programs running on the respective computers and having a client-server relationship with each other. Servers can be cloud servers, distributed system servers, or servers incorporating blockchain technology.

[0211] It should be understood that the various forms of processes shown above can be used to rearrange, add, or delete steps. For example, the steps described in this disclosure can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this disclosure can be achieved, and this is not limited herein.

[0212] The specific embodiments described above do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure should be included within the scope of protection of this disclosure.

Claims

1. An object recommendation method, comprising: Based on the first representation vector of the resource consumption object and the second representation vector set of the candidate resource production object set, a first similarity between the resource consumption object and the candidate resource production object set is determined, and at least one first similarity is obtained. The first representation vector is used to represent the object attention behavior of the resource consumption object in multiple applications, and the object attention behavior represents the attention behavior of the resource consumption object to the resource production object it is interested in. Based on the at least one first similarity, a target resource production object set is determined from the candidate resource production object set; and The target resource production object set is recommended to the resource consumer. Wherein, the first representation vector is obtained by processing the fifth comprehensive object attention behavior data of the resource consumption object node corresponding to the resource consumption object using the attention relationship model; the attention relationship model is obtained by training a graph neural network model using the attention relationship graph; the attention relationship graph is obtained based on the comprehensive object attention behavior dataset of the resource object set; the resource object set is determined from multiple candidate resource objects included in the candidate resource object set based on the sample screening strategy and the candidate object attention behavior dataset of the candidate resource object set; The sample screening strategy includes at least one of the following: A time-dimensional strategy is used to determine whether the timing of the candidate object attention behavior in the candidate object attention behavior dataset meets predetermined conditions. The resource consumption object dimension strategy is used to evaluate the level of attention behavior activity when the candidate resource object is a resource consumption object; The resource production object dimension strategy is used to evaluate the quality level of objects when the candidate resource object is a resource production object.

2. The method according to claim 1, further comprising: Based on the first representation vector of the resource consumption object and the third representation vector set of the resource consumption object set, a second similarity between the resource consumption object and the resource consumption object set is determined, and at least one second similarity is obtained. Based on the at least one second similarity, a candidate resource consumption object set is determined from the resource consumption object set; as well as Based on the candidate resource consumption object set, determine the candidate resource production object set.

3. The method according to claim 1, further comprising: Determine the first set of resource production objects corresponding to the resource consumption objects; Based on the fourth representation vector set of the first resource production object set and the second representation vector set of the second resource production object set, a third similarity between the first resource production object set and the second resource production object set is determined, and at least one third similarity is obtained. as well as The candidate resource production object set is determined from the second resource production object set based on the at least one third similarity.

4. The method according to any one of claims 1 to 3, wherein, The candidate resource production object set includes at least one candidate resource production object; Wherein, determining the target resource production object set from the candidate resource production object set based on the at least one first similarity includes: Based on the at least one first similarity, the at least one candidate resource production object is sorted to obtain first sorting information; and Based on the first sorting information, the target resource production object set is determined from the at least one candidate resource production object.

5. The method according to any one of claims 1 to 3, further comprising: Identification information matching the first identification information of the resource consumption object is determined from the set of association relationships; as well as The representation vector corresponding to the identification information that matches the first identification information is determined as the first representation vector; The set of associations includes at least one association, and the association represents the representation vector corresponding to the relationship between the identification information and the representation vector.

6. The method according to claim 5, further comprising: From the set of association relationships, determine the set of identifier information that matches the second set of identifier information of the candidate resource production object set; as well as The set of representation vectors corresponding to the set of identification information that matches the second set of identification information is determined as the second set of representation vectors.

7. The method according to claim 1, wherein, The first representation vector is determined based on the first integrated object's attention behavior data and the second integrated object's attention behavior data of the first neighbor resource production object; The first comprehensive object attention behavior data is determined based on multiple first object attention behavior data, which correspond to the multiple applications. The first neighbor resource production object is determined based on the attention behavior data of the first comprehensive object; The second comprehensive object attention behavior data is determined based on multiple second object attention behavior data, which correspond to the multiple applications.

8. The method according to claim 7, wherein, The first representation vector is determined based on the first fused feature data; Wherein, the first fused feature data is obtained by fusing the first aggregated feature data and the behavioral feature data corresponding to the first comprehensive object attention behavior data; The first fused feature data is obtained by aggregating features from the second comprehensive object attention behavior data.

9. The method according to claim 7 or 8, wherein, The first comprehensive object attention behavior data is obtained by processing the multiple first object attention behavior data using a predetermined format strategy, and the multiple first object attention behavior data have the same data format.

10. The method according to claim 1, wherein, The candidate resource production object set includes at least one candidate resource production object, and the second representation vector set includes a second representation vector corresponding to the at least one candidate resource production object. The second representation vector is determined based on the third integrated object's attention behavior data and the fourth integrated object's attention behavior data of the second neighbor resource production object; The third comprehensive object attention behavior data is determined based on multiple third object attention behavior data, which correspond to the multiple applications. The second neighbor resource production object is determined based on the attention behavior data of the third comprehensive object; The fourth comprehensive object attention behavior data is determined based on multiple fourth object attention behavior data, which correspond to the multiple applications.

11. The method according to claim 10, wherein, The second representation vector is obtained based on the second fused feature data; The second fused feature data is obtained by fusing the second aggregated feature data and the behavioral feature data corresponding to the third comprehensive object attention behavior data; The second aggregated feature data is obtained by aggregating the features of the fourth comprehensive object's attention behavior data.

12. The method according to claim 10 or 11, wherein, The third comprehensive object attention behavior data is obtained by processing the multiple third object attention behavior data using a predetermined format strategy, and the multiple third object attention behavior data have the same data format.

13. The method according to any one of claims 1 to 3, wherein, The candidate resource production object set includes at least one candidate resource production object, and the second representation vector set includes a second representation vector corresponding to the at least one candidate resource production object. The second representation vector is obtained by processing the sixth comprehensive object attention behavior data of the resource production object node corresponding to the candidate resource production object using the attention relationship model.

14. The method according to claim 13, wherein, in, The attention graph includes multiple resource object nodes and at least one edge. The multiple resource object nodes include at least one resource consumer object node and at least one resource producer object node. The edge is used to represent the attention behavior between the connected resource consumer object node and resource producer node. The resource object set includes at least one resource object, and the comprehensive object attention behavior dataset includes comprehensive object attention behavior data corresponding to the at least one resource object. The comprehensive object attention behavior data is determined based on multiple object attention behavior data, and the object attention behavior data represents the attention behavior data of the resource production object when the resource object is a resource consumption object.

15. The method according to claim 14, wherein, The multiple object attention behavior data are obtained using data interfaces corresponding to the multiple applications.

16. An object recommendation device, comprising: The first determining module is configured to determine a first similarity between the resource consumption object and the candidate resource production object set based on a first representation vector of the resource consumption object and a second representation vector set of the candidate resource production object set, thereby obtaining at least one first similarity, wherein the first representation vector is used to represent the object attention behavior of the resource consumption object in multiple applications, and the object attention behavior represents the attention behavior of the resource consumption object to the resource production object it is interested in; The second determining module is configured to determine a target resource production object set from the candidate resource production object set based on the at least one first similarity; and The recommendation module is used to recommend the target resource production object set to the resource consumer object; Wherein, the first representation vector is obtained by processing the fifth comprehensive object attention behavior data of the resource consumption object node corresponding to the resource consumption object using the attention relationship model; the attention relationship model is obtained by training a graph neural network model using the attention relationship graph; the attention relationship graph is obtained based on the comprehensive object attention behavior dataset of the resource object set; the resource object set is determined from multiple candidate resource objects included in the candidate resource object set based on the sample screening strategy and the candidate object attention behavior dataset of the candidate resource object set; The sample screening strategy includes at least one of the following: A time-dimensional strategy is used to determine whether the timing of the candidate object attention behavior in the candidate object attention behavior dataset meets predetermined conditions. The resource consumption object dimension strategy is used to evaluate the level of attention behavior activity when the candidate resource object is a resource consumption object; The resource production object dimension strategy is used to evaluate the quality level of objects when the candidate resource object is a resource production object.

17. An electronic device comprising: At least one processor; as well as A memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the method of any one of claims 1 to 15.

18. A non-transitory computer-readable storage medium storing computer instructions, wherein, The computer instructions are used to cause the computer to perform the method according to any one of claims 1 to 15.

19. A computer program product comprising a computer program that, when executed by a processor, implements the method according to any one of claims 1 to 15.

Images