A resource recommendation method and device, electronic equipment and storage medium

Abstract

The application relates to the technical field of artificial intelligence, in particular to a resource recommendation method and device, an electronic device and a storage medium, which are used to improve resource recommendation accuracy. The method comprises the following steps: acquiring an initial object network constructed based on a plurality of objects, the connection relationship between each two nodes in the network being determined based on object attribute features of the network; performing matrix conversion on an adjacency matrix corresponding to the initial object network based on a preset conversion strategy, and obtaining an intermediate object network based on the converted adjacency matrix; the number of nodes of the intermediate object network is less than that of the initial object network; performing at least one node merging processing on the intermediate object network based on a preset node merging strategy, and obtaining at least one target object network; performing community division based on at least one of the intermediate object network and the target object network, and performing resource recommendation according to the community division result. Based on the conversion and node merging strategies, the resource recommendation accuracy can be improved.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence technology, and in particular to a resource recommendation method, apparatus, electronic device, and storage medium. Background Technology

[0002] With the rapid development of network technology, terminal devices now have more diversified functions, such as watching programs and advertisements, shopping, and entertainment. In implementing these functions, intelligent recommendation systems can be used to recommend resources to users, enabling them to quickly locate resources of interest.

[0003] With the rapid development of artificial intelligence (AI) technology, intelligent recommendation systems are being applied in various fields. Taking smart TV (TV) recommendation systems as an example: when users use TV to shop online, the intelligent TV recommendation system can combine useful information mined from the user's network to recommend products to the user.

[0004] Intelligent TV recommendation systems in related technologies are based on community discovery techniques, which divide TV user networks into communities and then recommend resources. Specifically, by studying the characteristics of sub-communities within the user network, the system indirectly describes the network's attributes and uses community discovery techniques to obtain valuable community structure information, such as shopping habits. However, when performing the above processing based on related community discovery methods, network feature information is easily lost, leading to inaccurate resource recommendations.

[0005] Therefore, how to reduce the impact of inaccurate resource recommendations is an urgent issue to be addressed. Summary of the Invention

[0006] This application provides a resource recommendation method, apparatus, electronic device, and storage medium to improve the accuracy of resource recommendations. The resource recommendation method provided in this application includes:

[0007] Obtain an initial object network constructed from multiple objects, wherein each node in the initial object network corresponds to an object, and the connection relationship between any two nodes is determined based on the object attribute characteristics of the initial object network;

[0008] Based on a preset transformation strategy, the adjacency matrix corresponding to the initial object network is transformed, and based on the transformed adjacency matrix, an intermediate object network corresponding to the initial object network is obtained; the number of nodes in the intermediate object network is less than that in the initial object network.

[0009] Based on a preset node merging strategy, the intermediate object network is subjected to at least one node merging process to obtain at least one target object network corresponding to the initial object network; the number of nodes in the target object network is less than that in the intermediate object network.

[0010] Community segmentation is performed based on at least one of the intermediate object network and the target object network, and resource recommendations are made based on the community segmentation results.

[0011] Optionally, the object attribute features include: local features for characterizing attributes unique to an individual object, and global features for characterizing attributes shared between objects;

[0012] The connection relationships between every two nodes in the initial object network are determined in the following manner:

[0013] For any two objects, if it is determined, based on the local features or the global features, that there is at least one object attribute related between the two objects, then there is a connection between the nodes corresponding to the two objects.

[0014] Optionally, obtaining the intermediate object network corresponding to the initial object network based on the transformed adjacency matrix includes:

[0015] For each node in the initial object network, the following operations are performed: determine each target element corresponding to a node from the transformed adjacency matrix; merge the other nodes corresponding to the maximum value of each target element with the node; wherein, each target element represents the degree of association between the node and another connected node;

[0016] The network obtained by merging nodes in the initial object network is used as the intermediate object network.

[0017] Optionally, the preset node merging strategy is a common neighbor node merging strategy;

[0018] The method, based on a preset node merging strategy, performs at least one node merging process on the intermediate object network to obtain at least one target object network corresponding to the initial object network, including:

[0019] Based on the common neighbor node merging strategy, the intermediate object network is subjected to at least one node merging process, wherein each node merging process executes the following procedure:

[0020] Merge at least two nodes with common neighbors in the current object network to be merged, and use the merged object network as a target object network;

[0021] In the first node merging process, the current object network to be merged is the intermediate object network; in each subsequent node merging process, the current object network to be merged is the target object network obtained in the previous merge.

[0022] Optionally, the community division of the plurality of objects based on at least one of the intermediate object network and the target object network includes at least one of the following:

[0023] The intermediate object network is divided into communities;

[0024] Divide any target object's network into communities;

[0025] Perform at least one network restoration process on any target object network, and divide any candidate object network obtained from the restoration into communities.

[0026] Optionally, performing at least one network restoration process on any target object network includes:

[0027] Based on a deep learning model, at least one network reconstruction process is performed on the network of any target object, wherein each reconstruction process executes the following steps:

[0028] When the node representation of the current object network to be restored is used as the input of the deep learning model, the node representation output by the deep learning model is obtained, and the object network determined based on the output node representation is taken as a candidate object network.

[0029] In the first network restoration process, the object network to be restored is any one of the target object networks; in each subsequent network restoration process, the object network to be restored is the candidate object network obtained in the previous restoration.

[0030] This application provides a resource recommendation device, which includes:

[0031] A network acquisition unit is used to acquire an initial object network constructed based on multiple objects, wherein each node in the initial object network corresponds to an object, and the connection relationship between any two nodes is determined based on the object attribute characteristics of the initial object network.

[0032] The network transformation unit is used to perform matrix transformation on the adjacency matrix corresponding to the initial object network based on a preset transformation strategy, and obtain an intermediate object network corresponding to the initial object network based on the transformed adjacency matrix; the number of nodes in the intermediate object network is less than that in the initial object network.

[0033] A node merging unit is used to perform at least one node merging process on the intermediate object network based on a preset node merging strategy to obtain at least one target object network corresponding to the initial object network; the number of nodes in the target object network is less than that in the intermediate object network.

[0034] The resource recommendation unit is used to perform community segmentation based on at least one of the intermediate object network and the target object network, and to recommend resources based on the community segmentation results.

[0035] Optionally, the object attribute features include: local features for characterizing attributes unique to an individual object, and global features for characterizing attributes shared between objects;

[0036] The network acquisition unit is specifically used to determine the connection relationship between every two nodes in the initial object network in the following ways:

[0037] For any two objects, if it is determined, based on the local features or the global features, that there is at least one object attribute related between the two objects, then there is a connection between the nodes corresponding to the two objects.

[0038] Optionally, the network conversion unit is specifically used for:

[0039] For each node in the initial object network, the following operations are performed: determine each target element corresponding to a node from the transformed adjacency matrix; merge the other nodes corresponding to the maximum value of each target element with the node; wherein, each target element represents the degree of association between the node and another connected node;

[0040] The network obtained by merging nodes in the initial object network is used as the intermediate object network.

[0041] Optionally, the preset node merging strategy is a common neighbor node merging strategy;

[0042] The node merging unit is specifically used for:

[0043] Based on the common neighbor node merging strategy, the intermediate object network is subjected to at least one node merging process, wherein each node merging process executes the following procedure:

[0044] Merge at least two nodes with common neighbors in the current object network to be merged, and use the merged object network as a target object network;

[0045] In the first node merging process, the current object network to be merged is the intermediate object network; in each subsequent node merging process, the current object network to be merged is the target object network obtained in the previous merge.

[0046] Optionally, the resource recommendation unit is specifically used for at least one of the following:

[0047] The intermediate object network is divided into communities;

[0048] Divide any target object's network into communities;

[0049] Perform at least one network restoration process on any target object network, and divide any candidate object network obtained from the restoration into communities.

[0050] Optionally, the resource recommendation unit is specifically used to perform at least one network restoration process on any target object network.

[0051] Based on a deep learning model, at least one network reconstruction process is performed on the network of any target object, wherein each reconstruction process executes the following steps:

[0052] When the node representation of the current object network to be restored is used as the input of the deep learning model, the node representation output by the deep learning model is obtained, and the object network determined based on the output node representation is taken as a candidate object network.

[0053] In the first network restoration process, the object network to be restored is any one of the target object networks; in each subsequent network restoration process, the object network to be restored is the candidate object network obtained in the previous restoration.

[0054] An electronic device provided in this application includes a processor and a memory, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor performs the steps of any of the above-described resource recommendation methods.

[0055] This application provides a computer-readable storage medium including a computer program. When the computer program is run on an electronic device, the computer program is used to cause the electronic device to perform the steps of any of the above-described resource recommendation methods.

[0056] This application provides a computer program product, including a computer program stored in a computer-readable storage medium. When a processor of an electronic device reads the computer program from the computer-readable storage medium, the processor executes the computer program, causing the electronic device to perform the steps of any of the above-described resource recommendation methods.

[0057] The beneficial effects of this application are as follows:

[0058] This application provides a resource recommendation method, apparatus, electronic device, and storage medium. Considering that when adjacency matrices are typically used to represent data in an object network, the presence of numerous zeros can lead to computational difficulties, this application employs a data preprocessing method. The adjacency matrix corresponding to the initial object network is transformed, and based on the transformed adjacency matrix, an intermediate object network corresponding to the initial object network is obtained. This intermediate object network has fewer nodes than the initial object network. This method effectively processes the node data of the object network, not only solving the computational difficulties caused by the sparsity of the adjacency matrix but also reflecting the correlation between objects more specifically. Furthermore, this application further uses a node merging strategy to merge nodes in the intermediate object network to obtain a target object network corresponding to the initial object network. This gradually reduces the network size while largely preserving the object attribute characteristics, thereby reducing the loss of network information due to the reduction in network size. Based on the above processing, community detection methods are used to divide the intermediate or target object network, improving the accuracy of resource recommendation.

[0059] Other features and advantages of this application will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the application. The objectives and other advantages of this application may be realized and obtained by means of the structures particularly pointed out in the written description, claims, and drawings. Attached Figure Description

[0060] The accompanying drawings, which are provided to further illustrate this application and form part of this application, illustrate exemplary embodiments of this application and are used to explain this application, but do not constitute an undue limitation of this application.

[0061] Figure 1 This is a schematic diagram of the application scenario in the embodiments of this application;

[0062] Figure 2 This is a flowchart illustrating the implementation of a resource recommendation method in an embodiment of this application.

[0063] Figure 3This is a schematic diagram of an initial object network in an embodiment of this application;

[0064] Figure 4 This is a schematic diagram of an intermediate object network in an embodiment of this application;

[0065] Figure 5a This is a schematic diagram of a first target object network in an embodiment of this application;

[0066] Figure 5b This is a schematic diagram of a second target object network in an embodiment of this application;

[0067] Figure 6 This is a schematic diagram of a node merging embodiment in this application;

[0068] Figure 7 This is a schematic diagram of one community division method in an embodiment of this application;

[0069] Figure 8 This is a schematic diagram of a network restoration method according to an embodiment of this application;

[0070] Figure 9 This is a schematic diagram of a candidate object network in an embodiment of this application;

[0071] Figure 10a This is a schematic diagram illustrating the specific implementation process of a resource recommendation method in an embodiment of this application;

[0072] Figure 10b This is a schematic diagram illustrating the specific implementation process of a resource recommendation method in an embodiment of this application;

[0073] Figure 10c This is a schematic diagram illustrating the specific implementation process of a resource recommendation method in an embodiment of this application;

[0074] Figure 10d This is a schematic diagram illustrating the specific implementation process of a resource recommendation method in an embodiment of this application;

[0075] Figure 11 This is a schematic diagram of the composition structure of a resource recommendation device according to an embodiment of this application;

[0076] Figure 12 This is a schematic diagram of the composition structure of an electronic device according to an embodiment of this application;

[0077] Figure 13 This is a schematic diagram of the hardware structure of a computing device according to an embodiment of this application. Detailed Implementation

[0078] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of this application will be clearly and completely described below with reference to the accompanying drawings of the embodiments of this application. Obviously, the described embodiments are only some embodiments of the technical solutions of this application, and not all embodiments. Based on the embodiments recorded in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the technical solutions of this application.

[0079] The following describes some of the concepts involved in the embodiments of this application.

[0080] Resources: refers to information content that can be published, transmitted, and stored on the network, such as advertisements, products, videos, songs, etc.

[0081] Object networks and communities: An object network is a network composed of multiple nodes (objects). The connections between nodes in this network are determined based on whether there are certain relationships between the attributes of the corresponding objects. An object network consists of communities of different sizes. Given an object network, the subgraph formed by nodes with similar attributes is called a community. This application's embodiments involve an initial object network, an intermediate object network, a target object network, and a candidate object network. Specifically, based on a transformation strategy, an intermediate object network is obtained from the initial object network; based on a node merging strategy, a target object network is obtained from the intermediate object network; and based on network restoration, a candidate object network is obtained from the target object network.

[0082] Object attribute features include local features that characterize the unique attributes of an individual object, and global features that characterize the attributes shared between objects. For example, gender and unique hobbies are local features, while shared hobbies and mutual friends are global features.

[0083] Community discovery: Given a network of objects, the process of identifying communities. For example, consider the knowledge attributes of employees in an industrial park. Employees within the same company mostly know each other, while employees from different companies know each other less. Based on this knowledge attribute, each company constitutes a community, thus identifying these communities. In this application, this technology can be used to group an object in the object network and objects directly related to that object, or objects and objects with some similarity to that object, into the same community, thereby recommending similar resources.

[0084] Adjacency matrix: A matrix that represents the adjacency relationships between vertices. For example, for an object network containing N objects, an N×N adjacency matrix is ​​used. If two objects are related, the corresponding position in the adjacency matrix is ​​set to 1; otherwise, it is set to 0.

[0085] Transformation Strategy: Preset matrix transformation strategies are used to transform the adjacency matrix. For example, modularity functions, Euclidean distance functions, etc., can be used for matrix transformation. When the adjacency matrix contains a large number of zeros, which makes calculation difficult, matrix transformation can convert the zeros into non-zero data (such as decimals between 0 and 1), thus solving the computational difficulties caused by the sparsity due to the large number of zeros.

[0086] Node merging strategy: A preset node merging strategy is used to merge corresponding nodes with similar object attribute characteristics in the object network. For example, nodes with common neighbors can be merged.

[0087] The preferred embodiments of this application are described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit this application. Furthermore, the embodiments and features in the embodiments of this application can be combined with each other without conflict.

[0088] The resource recommendation method provided by the exemplary embodiments of this application will be described below with reference to the accompanying drawings and the application scenarios described above. It should be noted that the application scenarios described above are only shown to facilitate understanding of the spirit and principles of this application, and the embodiments of this application are not limited in any way in this respect.

[0089] See Figure 1 The diagram shown is an application scenario illustration of an embodiment of this application. The application scenario diagram includes two terminal devices 110 and one server 120.

[0090] In this embodiment, the terminal device 110 includes, but is not limited to, mobile phones, tablets, laptops, desktop computers, e-book readers, smart voice interaction devices, smart home appliances, and in-vehicle terminals. The terminal device may have a resource recommendation-related client installed. This client can be software (e.g., browsers, video software, shopping software, music software), or a webpage, mini-program, etc. The server 120 is the backend server corresponding to the software, webpage, or mini-program, or a server specifically used for resource recommendation; this application does not impose specific limitations. The server 120 can be an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDN), and big data and artificial intelligence platforms.

[0091] It should be noted that the resource recommendation method in each embodiment of this application can be executed by an electronic device, which can be a terminal device 110 or a server 120. That is, the method can be executed by the terminal device 110 or the server 120 alone, or by the terminal device 110 and the server 120 together.

[0092] In one alternative implementation, the terminal device 110 and the server 120 can communicate via a communication network.

[0093] In one alternative implementation, the communication network is a wired network or a wireless network.

[0094] It should be noted that, Figure 1 The examples shown are merely illustrative; in reality, the number of terminal devices and servers is unlimited and is not specifically limited in the embodiments of this application.

[0095] See Figure 2 The diagram shown is a flowchart of a resource recommendation method provided in this application embodiment. Taking a terminal device as the execution subject, the specific implementation process of this method includes the following S21-S24:

[0096] S21: The terminal device obtains an initial object network based on multiple objects.

[0097] In this system, each node in the initial object network corresponds to an object, and the connection between any two nodes is determined based on the object attribute characteristics of the initial object network.

[0098] In the embodiments of this application, the object (such as the user) differs depending on the application scenario. For example, in a TV network scenario, the object can be a user who has activated a TV subscription; in a short video network scenario, the object can be a user who has registered for the short video app; and in a shopping network scenario, the object can be a user who has registered for the shopping app. Correspondingly, the resources can be advertisements and programs recommended when a user watches TV; short videos and advertisements recommended when a user browses short video apps; or products recommended when making TV shopping, etc.

[0099] It should be noted that the TV network scenarios, short video network scenarios, and e-commerce network scenarios listed above are only simplified examples and are not specifically limited in this article. The following text mainly focuses on advertising recommendations in the TV network scenario for detailed explanation:

[0100] Specifically, the object attribute features of the initial object network represent the basic information-related attributes of each object in the network. Considering that there are many users in the object network, and some users may have certain similarities, an optional implementation method is that the object attribute features include, but are not limited to, local features and global features. Local features are used to represent individual object-specific attributes, such as gender and unique hobbies. Global features are used to represent shared attributes among objects, such as common hobbies and mutual friends.

[0101] See Figure 3 The diagram shown is a schematic of an initial object network in an embodiment of this application. The diagram contains 10 nodes, numbered 1-10. An edge exists between node 1 and node 2, indicating a connection between them; no edge exists between node 2 and node 3, indicating no connection between them; and so on.

[0102] It should be noted that the initial number of network nodes listed above as 10 is just an example and is not a specific limitation in this article.

[0103] In this application, when determining the connection relationship between every two nodes in the initial object network based on the object attribute characteristics of the initial object network, an optional implementation method is as follows:

[0104] For any two objects, if it is determined that there is at least one object attribute related between the two objects based on local or global features, then it indicates that there is a connection between the nodes corresponding to the two objects.

[0105] In the embodiments of this application, object attribute correlation indicates that there is a certain relationship between object attributes. Specifically, attribute correlation can mean that two objects have exactly the same attribute, for example, all users watching TV commercials are male; it can also mean that two objects have similar attributes, that is, their attributes are similar within a certain range. For example, all users watching TV commercials are young people, belonging to the same age range; and so on.

[0106] Still with Figure 3 As shown in the example, the connection between node 1 and node 2 is because both nodes 1 and 2, which are watching TV commercials, are young people.

[0107] In the above method, based on the characteristics of object attributes, it is possible to distinguish whether there is a relationship between the nodes corresponding to the objects.

[0108] S22: Based on a preset transformation strategy, the adjacency matrix corresponding to the initial object network is transformed, and based on the transformed adjacency matrix, an intermediate object network corresponding to the initial object network is obtained.

[0109] The intermediate object network has fewer nodes than the initial object network. The adjacency matrix of the initial object network can be denoted as A, and the transformed adjacency matrix can be denoted as A'.

[0110] In this embodiment, the preset transformation strategies include, but are not limited to, modularity function transformation and Euclidean distance function transformation. Taking the modularity function as an example, this function is a method for measuring the strength of network community structure and can have various calculation formulas. The following is an example of one modularity formula:

[0111]

[0112] Where Q represents the modularity function; for an undirected network, m represents the sum of the weights of all edges in the network, and for a directed network, it is half the sum of the weights of all edges in the network; A ij K represents the weight of the edge between node i and node j. i δ(C) represents the sum of the weights of all edges pointing to node i; i C j ) indicates that if node i and node j are in the same community, then δ(C) i C j If the result is 1, then the result is 0; otherwise, the result is 0.

[0113] In this embodiment, if node i and node j are related, their corresponding weight is 1; otherwise, it is 0. For example, taking node 1 and node 2 as an example, these two nodes are connected, and their corresponding weight is A. 12 The weight is 1; taking nodes 2 and 3 as examples, these two nodes have no connection relationship, and their corresponding weights are 1. 23 The value is 0; and so on.

[0114] It should be noted that the modularity function transformation strategies listed above are merely illustrative examples and are not intended to be specific in this paper. Furthermore, the object network in the embodiments of this application can be an undirected network or a directed network, and can be flexibly set according to the actual application scenario; no specific limitations are imposed in this paper.

[0115] Optionally, for each node in the initial object network, perform the following operations: determine the target elements corresponding to a node from the transformed adjacency matrix; merge the other nodes corresponding to the maximum values ​​of the target elements with a node; and then use the network after merging the nodes of the initial object network as the intermediate object network.

[0116] Each target element represents the degree of association between a node and another connected node.

[0117] In this embodiment, the association tightness characterizes the strength of the connection between nodes in the object network. The greater the association tightness, the stronger the connection between nodes.

[0118] Based on Q in the above formula, A can be transformed into A', where the data in A' are all decimals between 0 and 1, while the data in A are all 0 or 1. This transformation avoids the sparsity problem caused by the adjacency matrix A. Let's continue with... Figure 3 As shown in the example, the weight between node 1 and node 2 is 1, the weight between node 1 and node 3 is 1, and the weight between node 1 and node 4 is 1. Assuming that the association tightness A' between node 1 and node 2 can be obtained by calculating using the modularity function formula, 12 =0.8, the affinity A' between node 1 and node 3 13 =0.7, the affinity A' between node 1 and node 4 14 =0.6. Based on this, the correlation between node 1 and node 2 is greater than that between node 1 and node 3, and also greater than that between node 1 and node 4. The connection between node 1 and node 2 is stronger.

[0119] In this embodiment, node merging represents merging a node with its associated nodes by selecting the node with the strongest connection to that node. Based on this, node 1 and node 2 are merged to form a new node, which can be called a supernode. In the above method, nodes can be merged into a supernode using Q, thereby reducing the network size.

[0120] See Figure 4 The diagram shown is a schematic diagram of an intermediate object network in an embodiment of this application. The intermediate object network is obtained by transforming the initial object network based on the modularity function. In the diagram, node 1' is obtained by merging node 1 and node 2, node 5' is obtained by merging node 5 and node 10, node 6' is obtained by merging node 6 and node 7, and node 8' is obtained by merging node 8 and node 9.

[0121] In the above method, based on a preset transformation strategy, the adjacency matrix corresponding to the initial object network is transformed, and based on the transformed adjacency matrix, an intermediate object network is obtained. This solves the computational difficulties caused by the sparsity due to a large number of zeros, and can also reflect the correlation between objects.

[0122] S23: Based on a preset node merging strategy, perform at least one node merging process on the intermediate object network to obtain at least one target object network corresponding to the initial object network.

[0123] The target object network has fewer nodes than the intermediate object network.

[0124] In this embodiment, the preset node merging strategy includes, but is not limited to, the modularity gain strategy and the custom merging strategy. The custom merging strategy is a node merging strategy set according to actual needs. This paper takes the common neighbor node merging strategy as an example:

[0125] If node i and node j both have the same neighbor p, then node i and node j can be merged into one node.

[0126] Optionally, based on the common neighbor node merging strategies listed above, when performing at least one node merging process on the intermediate object network to obtain at least one target object network corresponding to the initial object network, each node merging process can execute the following procedure:

[0127] Merge at least two nodes with common neighbors in the current object network to be merged, and use the merged object network as a target object network.

[0128] In the first node merging process, the current object network to be merged is the intermediate object network; in each subsequent node merging process, the current object network to be merged is the target object network obtained from the previous merge.

[0129] Still with Figure 4 As shown in the example, node 1' and node 3 share a common neighbor, node 4. Node 1' and node 3 are merged to obtain node 13'. For Figure 4 The intermediate object network shown above, based on the common neighbor nodes listed above, can be used to obtain the target object network after node merging.

[0130] See Figure 5a As shown, it is a schematic diagram of the first type of target object network in the embodiment of this application, wherein the target object network G1 is obtained after node 1' and node 3 in the intermediate object network G' undergo a node merging.

[0131] exist Figure 5a Based on this, the first target object network undergoes a node merging process to obtain the second target object network. (See [link / reference]). Figure 5b As shown, it is a schematic diagram of the second type of target object network in the embodiment of this application, wherein the target object network G2 is obtained after node 6' and node 8' in the target object network G1 are merged once.

[0132] Specifically, in the embodiments of this application, the target object network representation is obtained from the intermediate object network through a preset node merging strategy. By merging the intermediate object network multiple times, a series of target object networks G1, G2, G3, ..., G n, (G1>G2>...>G i ...>G n In this process, the intermediate object network is denoted by G'. G' undergoes one node merging to obtain the target object network G1; G1 undergoes one node merging to obtain the target object network G2; G2 undergoes one node merging to obtain the target object network G3, and so on.

[0133] See Figure 6 The diagram shown is a schematic representation of node merging in an embodiment of this application. Taking the network to be merged as G' as an example, Figure 6 The process of obtaining the target network by merging nodes G' four times is illustrated. G' is merged into G1 after the first node merge; G1 is merged into G2 after the second node merge; G2 is merged into G3 after the third node merge; and G3 is merged into G4 after the fourth node merge. Based on this, the network size gradually shrinks after multiple node merges.

[0134] It should be noted that the target object networks listed above are only examples and are not specifically limited in this article.

[0135] In the above method, based on a preset node merging strategy, the intermediate network is processed by merging nodes to obtain the target object network, so that the network size is gradually reduced, and the object attribute characteristics are largely preserved, reducing the loss of network information caused by the reduction of network size.

[0136] S24: Based on at least one of the intermediate object network and the target object network, perform community segmentation, and recommend resources based on the community segmentation results.

[0137] In this embodiment, the community segmentation representation divides nodes in an object network into communities based on object attribute characteristics. For example, in a TV network object, the object can be divided into youth communities, middle-aged communities, etc.

[0138] Optionally, communities are divided among multiple objects based on at least one of the intermediate object network and the target object network, including at least one of the following:

[0139] Object network selection method one: Divide the intermediate object network into communities;

[0140] If the number of objects in the object network is large and falls within the second quantity range, the intermediate object network can be divided into communities.

[0141] Object network selection method two: Divide any target object network into communities;

[0142] If the number of objects in the object network is larger and falls within the third quantity range, the target object network can be divided into communities.

[0143] The third method for selecting object networks involves performing at least one network restoration process on any target object network and then dividing any candidate object network obtained from the restoration into communities.

[0144] If the number of objects in the object network is extremely large, falling within the fourth quantity range, community partitioning can be performed on the candidate object network.

[0145] When dividing any object network into communities, the following rules can be adopted: Taking the TV object network as an example, communities are divided according to whether they have the same interest and hobby attributes. Three communities are divided: Community 1 represents users who like to watch sports programs, Community 2 represents users who like to watch children's programs, and Community 3 represents users who like to watch TV commercials.

[0146] It should be noted that the above-listed object network community divisions are just simple examples and are not specifically limited in this article.

[0147] See Figure 7 As shown, it is a schematic diagram of a community division method in an embodiment of this application. In the initial method, that is, when the number of objects in the object network is small and in the first quantity range, the initial object network is directly divided into communities.

[0148] The first to fourth quantity ranges listed above increase sequentially. For example, the first quantity range represents less than 1,000; the second quantity range represents 1,000 to 10,000; the third quantity range represents 10,000 to 1,000,000; and the fourth quantity range represents more than 1,000,000.

[0149] It should be noted that the above division of quantity ranges is just a simple example. Any method of dividing the ranges is applicable to the embodiments of this application, and will not be described in detail here.

[0150] Optionally, when performing network reconstruction processing on any target object network at least once, it can be implemented based on a deep learning model:

[0151] When performing at least one network reconstruction process on any target object network based on a deep learning model, each reconstruction process can execute the following steps:

[0152] When the node representation of the current object network to be restored is used as input to the deep learning model, the node representation output by the deep learning model is obtained, and the object network determined based on the output node representation is used as a candidate object network.

[0153] In the first network restoration process, the object network to be restored is any target object network; in each subsequent network restoration process, the object network to be restored is the candidate object network obtained in the previous restoration.

[0154] In this embodiment, the network reconstruction processing uses a deep learning model to reconstruct the target object network into the upper-layer network, i.e., the candidate object network. The deep learning model, taking the GCN model as an example:

[0155] G i The representation of a layer network node is used as the G'th layer network node. i-1 The pre-representation of the layer network nodes, after being processed by the GCN model, yields G'. i-1 The true representation of the layer network nodes, further, will be G' i-1 The node representation of the layer network is as G' i-2 Pre-representation of layer network nodes.

[0156] See Figure 8 As shown, this is a schematic diagram of network restoration in an embodiment of this application, taking the target network G4 to be restored as an example. Figure 8 The process of obtaining the corresponding candidate object network by performing three network reductions on G4 is described. After the first network reduction, G'3 is obtained; after the second network reduction, G'2 is obtained; and after the third network reduction, G'1 is obtained. After multiple network reductions, the size of the object network gradually increases.

[0157] It should be noted that the candidate networks listed above are merely illustrative examples and are not intended to be specific in this article.

[0158] Specifically, still in Figure 5b Taking the example shown, the target object network G2 is subjected to network reconstruction processing. The representation of the network nodes in layer G2 is used as the pre-representation of the network nodes in layer G'1. After processing by the GCN model, the true representation of the network nodes in layer G'1 is obtained. During reconstruction, the result may be exactly the same as that of the target object network in that layer, or there may be some differences, but they are basically similar.

[0159] refer to Figure 9 As shown, taking the case where the restored result is exactly the same as the target object network of the same layer as the target object network G2, the candidate object network G'1 obtained by restoring the target object network G2 is exactly the same as the target object network G1 of the same layer.

[0160] If the initial object network contains a large number of objects and the relevant technologies cannot obtain the object feature information of the object network, a series of smaller target object networks can be obtained based on the node merging strategy. The object attribute feature information of the smaller target object networks can then be obtained. In this way, a series of network restoration processes can be performed to obtain the object attribute feature information of the candidate object networks.

[0161] In this application embodiment, different object networks can be selected for community division according to different actual application scenarios, so as to obtain the optimal division result.

[0162] The following is combined with Figure 7 The following are the resource recommendation processes for each of the listed methods:

[0163] If the number of objects in the object network is small, falling within the first quantity range, the initial object network can be directly divided into communities, i.e., using... Figure 7 The initial method shown.

[0164] for Figure 7 The initial method shown is described in the reference. Figure 10a The diagram shown illustrates a specific implementation flow of a resource recommendation method according to an embodiment of this application. The specific implementation flow of this method is as follows:

[0165] Step S1001a: Obtain the initial object network;

[0166] Step S1002a: Based on the community partitioning module, directly perform community partitioning on the initial object network;

[0167] Step S1003a: Based on the intelligent recommendation module, resource recommendations are made according to the corresponding community division results.

[0168] If the number of objects in the object network is large, falling within the second quantity range, community partitioning can be performed on the intermediate object network, i.e., using... Figure 7 Method 1 is shown.

[0169] for Figure 7 Method 1 shown, please refer to Figure 10b The diagram shown illustrates a specific implementation flow of a resource recommendation method according to an embodiment of this application. The specific implementation flow of this method is as follows:

[0170] Step S1001b: Obtain the initial object network;

[0171] Step S1002b: In the initial object network preprocessing module, based on the preset transformation strategy, the adjacency matrix corresponding to the initial object network is transformed, and the intermediate object network is obtained based on the transformed adjacency matrix.

[0172] Step S1003b: Based on the community partitioning module, perform community partitioning on the intermediate object network;

[0173] Step S1004b: Based on the intelligent recommendation module, resource recommendations are made according to the corresponding community division results.

[0174] If the number of objects in the object network is larger, falling within the third quantity range, community partitioning can be performed on the target object network, i.e., using... Figure 7 Method 2 is shown.

[0175] for Figure 7 Method 2 shown, please refer to Figure 10c The diagram shown illustrates a specific implementation flow of a resource recommendation method according to an embodiment of this application. The specific implementation flow of this method is as follows:

[0176] Step S1001c: Obtain the initial object network;

[0177] Step S1002c: In the initial object network preprocessing module, based on the preset transformation strategy, the adjacency matrix corresponding to the initial object network is transformed, and the intermediate object network is obtained based on the transformed adjacency matrix.

[0178] Step S1003c: Based on a preset node merging strategy, perform at least one node merging process on the intermediate object network to obtain at least one target object network;

[0179] Step S1004c: Based on the community partitioning module, perform community partitioning on any target object network;

[0180] Step S1005c: Based on the intelligent recommendation module, resource recommendations are made according to the corresponding community division results.

[0181] If the number of objects in the object network is extremely large, falling within the fourth quantity range, community partitioning can be performed on the candidate object network, i.e., using... Figure 7 Method 3 is shown.

[0182] for Figure 7 Method 3 is shown below. Figure 10d The diagram shown illustrates a specific implementation flow of a resource recommendation method according to an embodiment of this application. The specific implementation flow of this method is as follows:

[0183] Step S1001d: Obtain the initial object network;

[0184] Step S1002d: In the initial object network preprocessing module, based on the preset transformation strategy, the adjacency matrix corresponding to the initial object network is transformed, and the intermediate object network is obtained based on the transformed adjacency matrix.

[0185] Step S1003d: Based on the preset node merging strategy, perform at least one node merging process on the intermediate object network to obtain at least one target object network;

[0186] Step S1004d: Perform network restoration processing on any target object network at least once to obtain at least one candidate object network;

[0187] Step S1005d: Based on the community partitioning module, perform community partitioning on any candidate object network;

[0188] Step S1006d: Based on the intelligent recommendation module, resource recommendations are made according to the corresponding community division results.

[0189] In the embodiments of this application, the accuracy of resource recommendations can be effectively improved based on the above-described method.

[0190] Based on the same inventive concept, embodiments of this application also provide a resource recommendation device. For example... Figure 11 As shown, this is a schematic diagram of the resource recommendation device 1100, which may include:

[0191] The network acquisition unit 1101 is used to acquire an initial object network constructed based on multiple objects. Each node in the initial object network corresponds to an object, and the connection relationship between any two nodes is determined based on the object attribute characteristics of the initial object network.

[0192] The network transformation unit 1102 is used to perform matrix transformation on the adjacency matrix corresponding to the initial object network based on a preset transformation strategy, and obtain an intermediate object network corresponding to the initial object network based on the transformed adjacency matrix; the number of nodes in the intermediate object network is less than that in the initial object network.

[0193] The node merging unit 1103 is used to perform at least one node merging process on the intermediate object network based on a preset node merging strategy to obtain at least one target object network corresponding to the initial object network; the number of nodes in the target object network is less than that in the intermediate object network.

[0194] The resource recommendation unit 1104 is used to perform community segmentation based on at least one of the intermediate object network and the target object network, and to recommend resources based on the community segmentation results.

[0195] Optionally, object attribute features include: local features used to characterize attributes unique to an individual object, and global features used to characterize attributes shared between objects;

[0196] The network acquisition unit 1101 is specifically used to determine the connection relationship between every two nodes in the initial object network in the following ways:

[0197] For any two objects, if it is determined that there is at least one object attribute related between the two objects based on local or global features, then the nodes corresponding to the two objects have a connection relationship.

[0198] Optionally, the network conversion unit 1102 is specifically used for:

[0199] For each node in the initial object network, perform the following operations: determine the target elements corresponding to a node from the transformed adjacency matrix; merge the other nodes corresponding to the maximum values ​​of the target elements with a node; where each target element represents the degree of association between a node and another connected node.

[0200] The network obtained by merging nodes in the initial object network will be used as the intermediate object network.

[0201] Optionally, the preset node merging strategy is the common neighbor node merging strategy;

[0202] The node merging unit 1103 is specifically used for:

[0203] Based on the common neighbor node merging strategy, at least one node merging process is performed on the intermediate object network, wherein each node merging process executes the following procedure:

[0204] Merge at least two nodes with common neighbors in the current object network to be merged, and use the merged object network as a target object network;

[0205] In the first node merging process, the current object network to be merged is the intermediate object network; in each subsequent node merging process, the current object network to be merged is the target object network obtained from the previous merge.

[0206] Optionally, the resource recommendation unit 1104 is specifically used for at least one of the following:

[0207] Divide the intermediate object network into communities;

[0208] Divide any target object's network into communities;

[0209] Perform at least one network restoration process on any target object network, and divide any candidate object network obtained from the restoration into communities.

[0210] Optionally, at least one network restoration process is performed on any target object network. The resource recommendation unit 1104 is specifically used for:

[0211] Based on a deep learning model, at least one network reconstruction process is performed on any target object network, wherein each reconstruction process executes the following steps:

[0212] When the node representation of the current object network to be restored is used as the input of the deep learning model, the node representation output by the deep learning model is obtained, and the object network determined based on the output node representation is used as a candidate object network.

[0213] In the first network restoration process, the object network to be restored is any target object network; in each subsequent network restoration process, the object network to be restored is the candidate object network obtained in the previous restoration.

[0214] For ease of description, the above sections are divided into modules (or units) according to their functions and described separately. Of course, in implementing this application, the functions of each module (or unit) can be implemented in one or more software or hardware components.

[0215] Having introduced the resource recommendation method and apparatus according to exemplary embodiments of this application, we will now introduce an electronic device according to another exemplary embodiment of this application.

[0216] Those skilled in the art will understand that various aspects of this application can be implemented as a system, method, or program product. Therefore, various aspects of this application can be specifically implemented in the following forms: a completely hardware implementation, a completely software implementation (including firmware, microcode, etc.), or a combination of hardware and software implementations, collectively referred to herein as a "circuit," "module," or "system."

[0217] Based on the same inventive concept as the above-described method embodiments, this application also provides an electronic device. In this embodiment, the structure of the electronic device can be as follows: Figure 12 As shown, it includes a memory 1201, a communication module 1203, and one or more processors 1202.

[0218] The memory 1201 is used to store computer programs executed by the processor 1202. The memory 1201 may mainly include a program storage area and a data storage area. The program storage area may store the operating system and programs required to run instant messaging functions, etc.; the data storage area may store various instant messaging information and operation instruction sets, etc.

[0219] Memory 1201 may be volatile memory, such as random-access memory (RAM); memory 1201 may also be non-volatile memory, such as read-only memory, flash memory, hard disk drive (HDD), or solid-state drive (SSD); or memory 1201 may be any other medium capable of carrying or storing a desired computer program having the form of instructions or data structures and accessible by a computer, but is not limited thereto. Memory 1201 may be a combination of the above-described memories.

[0220] Processor 1202 may include one or more central processing units (CPUs) or digital processing units, etc. Processor 1202 is used to implement the above-described register application information generation method when calling computer programs stored in memory 1201.

[0221] The communication module 1203 is used to communicate with terminal devices and other servers.

[0222] This application embodiment does not limit the specific connection medium between the memory 1201, communication module 1203, and processor 1202. This application embodiment... Figure 12 The memory 1201 and the processor 1202 are connected via a bus 1204, and the bus 1204 is in Figure 12 The diagram uses thick lines to describe the connections between other components; these are for illustrative purposes only and should not be considered limiting. The 1204 bus can be divided into address bus, data bus, control bus, etc. For ease of description, Figure 12 It is described using only a thick line, but does not indicate that there is only one bus or one type of bus.

[0223] The memory 1201 stores a computer storage medium, which stores computer-executable instructions for implementing the resource recommendation method of this application embodiment. The processor 1202 is used to execute the above-described resource recommendation method, such as... Figure 2 As shown.

[0224] The following reference Figure 13 To describe a computing device 1300 according to this embodiment of the present application. Figure 13 The computing device 1300 is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of this application.

[0225] like Figure 13The computing device 1300 is manifested in the form of a general-purpose computing device. The components of the computing device 1300 may include, but are not limited to: at least one processing unit 1301, at least one storage unit 1302, and a bus 1303 connecting different system components (including storage unit 1302 and processing unit 1301).

[0226] Bus 1303 represents one or more of several bus structures, including a memory bus or memory controller, peripheral bus, processor, or a local bus using any of the various bus structures.

[0227] Storage unit 1302 may include a readable medium in the form of volatile memory, such as random access memory (RAM) 1321 and / or cache memory 1322, and may further include read-only memory (ROM) 1323.

[0228] Storage unit 1302 may also include a program / utility 1325 having a set (at least one) of program modules 1324, such program modules 1324 including but not limited to: an operating system, one or more application programs, other program modules and program data, each or some combination of these examples may include an implementation of a network environment.

[0229] The computing device 1300 can also communicate with one or more external devices 1304 (e.g., keyboard, pointing device, etc.), and with one or more devices that enable a user to interact with the computing device 1300, and / or with any device that enables the computing device 1300 to communicate with one or more other computing devices (e.g., router, modem, etc.). This communication can be performed via the input / output (I / O) interface 1305. Furthermore, the computing device 1300 can also communicate with one or more networks (e.g., local area network (LAN), wide area network (WAN), and / or public networks, such as the Internet) via the network adapter 1306. Figure 13 As shown, network adapter 1306 communicates with other modules used in computing device 1300 via bus 1303. It should be understood that, although not shown in the figure, other hardware and / or software modules may be used in conjunction with computing device 1300, including but not limited to: microcode, device drivers, redundant processors, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0230] In some possible implementations, various aspects of the resource recommendation method provided in this application can also be implemented in the form of a program product, which includes a computer program. When the program product is run on an electronic device, the computer program causes the electronic device to perform the steps in the resource recommendation method according to the various exemplary embodiments of this application described above. For example, the electronic device can perform actions such as... Figure 2 The steps are shown in the figure.

[0231] The program product may employ any combination of one or more readable media. A readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples (a non-exhaustive list) of readable storage media include: electrical connections having one or more wires, portable 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 thereof.

[0232] The program product of the embodiments of this application may employ a portable compact disc read-only memory (CD-ROM) and include a computer program, and may run on a computing device. However, the program product of this application is not limited thereto. In this document, the readable storage medium may be any tangible medium that contains or stores a program that may be used by or in conjunction with a command execution system, apparatus, or device.

[0233] A readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, carrying a readable computer program. This propagated data signal may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A readable signal medium may also be any readable medium other than a readable storage medium, capable of sending, propagating, or transmitting a program for use by or in conjunction with a command execution system, apparatus, or device.

[0234] Computer programs contained on readable media may be transmitted using any suitable medium, including but not limited to wireless, wired, optical fiber, RF, etc., or any suitable combination thereof.

[0235] Computer programs for performing the operations of this application can be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java and C++, and conventional procedural programming languages ​​such as C or similar languages. The computer program can execute entirely on the user's computing device, partially on the user's device, as a standalone software package, partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server. In cases involving remote computing devices, the remote computing device can be connected to the user's computing device via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computing device (e.g., via the Internet using an Internet service provider).

[0236] It should be noted that although several units or sub-units of the device have been mentioned in the detailed description above, this division is merely exemplary and not mandatory. In fact, according to embodiments of this application, the features and functions of two or more units described above can be embodied in one unit. Conversely, the features and functions of one unit described above can be further divided and embodied by multiple units.

[0237] Furthermore, although the operations of the method of this application are described in a specific order in the accompanying drawings, this does not require or imply that these operations must be performed in that specific order, or that all the operations shown must be performed to achieve the desired result. Additionally or alternatively, certain steps may be omitted, multiple steps may be combined into one step, and / or one step may be broken down into multiple steps.

[0238] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing a computer-usable computer program.

[0239] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, produce a machine for implementing the flowchart illustrations. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0240] These computer program commands may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the commands stored in the computer-readable storage medium produce an article of manufacture including command means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0241] These computer program commands can also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing the commands executed on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0242] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.

[0243] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A resource recommendation method, characterized in that, The method includes: Obtain an initial object network constructed from multiple objects, wherein each node in the initial object network corresponds to an object, and the connection relationship between any two nodes is determined based on the object attribute characteristics of the initial object network; Based on a preset transformation strategy, the adjacency matrix corresponding to the initial object network is transformed, and based on the transformed adjacency matrix, an intermediate object network corresponding to the initial object network is obtained; wherein, the transformation strategy is based on the calculation of the affinity between nodes, and the number of nodes in the intermediate object network is less than that in the initial object network. Based on a preset node merging strategy, the intermediate object network is subjected to at least one node merging process to obtain at least one target object network corresponding to the initial object network; wherein, the node merging strategy is based on merging nodes with common neighbors, and the number of nodes in the target object network is less than that in the intermediate object network. Based on the number of objects in the initial object network, community partitioning is performed on at least one of the intermediate object network and the target object network, and resource recommendation is made based on the community partitioning results. Specifically, if the number of objects in the initial object network is in a second range, then the intermediate object network is partitioned into communities; if the number of objects in the initial object network is in a third range, then any target object network is partitioned into communities; if the number of objects in the initial object network is in a fourth range, then any target object network undergoes at least one network restoration process, and any candidate object network obtained from the restoration is partitioned into communities.

2. The method as described in claim 1, characterized in that, The object attribute features include: local features used to characterize the unique attributes of an individual object, and global features used to characterize the shared attributes among objects; The connection relationships between every two nodes in the initial object network are determined in the following manner: For any two objects, if it is determined, based on the local features or the global features, that there is at least one object attribute related between the two objects, then there is a connection between the nodes corresponding to the two objects.

3. The method as described in claim 1, characterized in that, The step of obtaining the intermediate object network corresponding to the initial object network based on the transformed adjacency matrix includes: For each node in the initial object network, the following operations are performed: determine each target element corresponding to a node from the transformed adjacency matrix; merge the other nodes corresponding to the maximum value of each target element with the node; wherein, each target element represents the degree of association between the node and another connected node; The network obtained by merging nodes in the initial object network is used as the intermediate object network.

4. The method as described in claim 1, characterized in that, The preset node merging strategy is a common neighbor node merging strategy; The method, based on a preset node merging strategy, performs at least one node merging process on the intermediate object network to obtain at least one target object network corresponding to the initial object network, including: Based on the common neighbor node merging strategy, the intermediate object network is subjected to at least one node merging process, wherein each node merging process executes the following procedure: Merge at least two nodes with common neighbors in the current object network to be merged, and use the merged object network as a target object network; In the first node merging process, the current object network to be merged is the intermediate object network; in each subsequent node merging process, the current object network to be merged is the target object network obtained in the previous merge.

5. The method as described in claim 1, characterized in that, The step of performing at least one network restoration process on any target object network includes: Based on a deep learning model, at least one network reconstruction process is performed on the network of any target object, wherein each reconstruction process executes the following steps: When the node representation of the current object network to be restored is used as the input of the deep learning model, the node representation output by the deep learning model is obtained, and the object network determined based on the output node representation is taken as a candidate object network. In the first network restoration process, the object network to be restored is any one of the target object networks; in each subsequent network restoration process, the object network to be restored is the candidate object network obtained in the previous restoration.

6. A resource recommendation device, characterized in that, The device includes: A network acquisition unit is used to acquire an initial object network constructed based on multiple objects, wherein each node in the initial object network corresponds to an object, and the connection relationship between any two nodes is determined based on the object attribute characteristics of the initial object network. A network transformation unit is used to transform the adjacency matrix corresponding to the initial object network based on a preset transformation strategy, and obtain an intermediate object network corresponding to the initial object network based on the transformed adjacency matrix; wherein the transformation strategy is based on the calculation of the correlation density between nodes, and the number of nodes in the intermediate object network is less than that in the initial object network. A node merging unit is used to perform at least one node merging process on the intermediate object network based on a preset node merging strategy to obtain at least one target object network corresponding to the initial object network; wherein the node merging strategy is based on merging nodes with common neighbors, and the number of nodes in the target object network is less than that in the intermediate object network. The resource recommendation unit is used to perform community partitioning based on at least one of the intermediate object network and the target object network, according to the object quantity size of the initial object network, and to recommend resources based on the community partitioning results; wherein, if the object quantity in the initial object network is within a second quantity range, the intermediate object network is partitioned into communities; if the object quantity in the initial object network is within a third quantity range, any target object network is partitioned into communities; if the object quantity in the initial object network is within a fourth quantity range, any target object network is subjected to at least one network restoration process, and any candidate object network obtained from the restoration is partitioned into communities.

7. An electronic device, characterized in that, It includes a processor and a memory, wherein the memory stores a computer program that, when executed by the processor, causes the processor to perform the steps of any one of the methods described in claims 1 to 5.

8. A computer-readable storage medium, characterized in that, It includes a computer program that, when run on an electronic device, causes the electronic device to perform the steps of any of the methods described in claims 1 to 5.

9. A computer program product, characterized in that, The method includes a computer program stored in a computer-readable storage medium; when a processor of an electronic device reads the computer program from the computer-readable storage medium, the processor executes the computer program, causing the electronic device to perform the steps of any one of claims 1 to 5.

Images