Group type identification method and apparatus, computer device, and medium

Abstract

Embodiments of the application disclose a group type identification method and device, computer equipment and a medium, and belong to the technical field of computers. The method comprises: acquiring first graph structure features of a first target graph and first user features of a plurality of first user nodes in the first target graph; processing the first graph structure features and the plurality of first user features to obtain an attention parameter corresponding to each first user node in the first target graph; selecting a plurality of second user nodes from the plurality of first user nodes according to the obtained plurality of attention parameters; and identifying first user features of the plurality of second user nodes and second graph structure features of a second target graph to obtain a group type of the target group. The method identifies the target group according to the user features and the graph structure features of the important user nodes selected, and improves the identification accuracy.

Description

TECHNICAL FIELD

[0001] Embodiments of the present application relate to the technical field of computer technology, and particularly relate to a group type identification method and device, a computer device, and a medium. BACKGROUND

[0002] With the development of computer technology and instant messaging applications, various types of groups can be established based on instant messaging applications, such as family groups, work groups, food groups, and the like. How to identify the group type has become a technical problem to be solved.

[0003] In the related art, the type of a group is determined according to user information corresponding to a plurality of user identifiers in the group. The user information includes user avatars, user tags, ages, genders, and the like. However, this method of determining the group type based on user information has a small amount of data and low accuracy of the determined group type. SUMMARY

[0004] Embodiments of the present application provide a group type identification method, device, computer device, and medium, which improves the accuracy of group type identification. The technical solution is as follows:

[0005] In one aspect, a group type identification method is provided, which includes:

[0006] obtaining first graph structure features of a first target graph and first user features of a plurality of first user nodes in the first target graph, the first user nodes being nodes corresponding to user identifiers in the target group, and the first target graph being constructed according to association relationships between the plurality of first user nodes;

[0007] processing the first graph structure features and the plurality of first user features to obtain attention parameters corresponding to each first user node in the first target graph, the attention parameters being used to represent the importance of the first user node in the first target graph;

[0008] selecting a plurality of second user nodes from the plurality of first user nodes according to the obtained plurality of attention parameters, the plurality of second user nodes corresponding to attention parameters greater than those of other first user nodes;

[0009] identifying first user features of the plurality of second user nodes and second graph structure features of a second target graph to obtain a group type of the target group, the second target graph being constructed according to association relationships between the plurality of second user nodes.

[0010] In a possible implementation, the graph structure feature comprises an association degree feature between any two user nodes in the plurality of user nodes, and the obtaining of the graph structure feature of the first target graph and the first user feature of the plurality of first user nodes in the first target graph comprises:

[0011] obtaining a co-occurrence frequency of any two user identifiers in the target group, the co-occurrence frequency being a number of times that the any two user identifiers jointly publish content in the target group within a plurality of reference time periods;

[0012] determining an association degree feature between the any two user identifiers according to the co-occurrence frequency, the association degree feature being in a positive correlation with the co-occurrence frequency.

[0013] In another possible implementation, the selecting of the plurality of second user nodes from the plurality of first user nodes according to the obtained plurality of attention parameters comprises:

[0014] multiplying a number of the plurality of first user nodes by a reference ratio to obtain a reference number;

[0015] arranging the attention parameters of the plurality of first user nodes in a descending order, selecting a reference number of attention parameters arranged in front, and taking the first user nodes corresponding to the selected attention parameters as the selected plurality of second user nodes.

[0016] In another possible implementation, the user feature comprises a user behavior feature and a user attribute feature, and the obtaining of the graph structure feature of the first target graph and the first user feature of the plurality of first user nodes in the first target graph comprises:

[0017] obtaining a user social network, the user social network comprising a plurality of registered user identifiers;

[0018] obtaining a user behavior feature of the plurality of user identifiers in the target group according to the user social network;

[0019] obtaining a user attribute feature of the plurality of user identifiers in the target group according to user portrait information corresponding to the plurality of user identifiers in the target group.

[0020] In another aspect, a group type identification apparatus is provided, and the apparatus comprises:

[0021] a feature obtaining module, configured to obtain a first graph structure feature of a first target graph and a first user feature of a plurality of first user nodes in the first target graph, the first user nodes being nodes corresponding to user identifiers in the target group, and the first target graph being constructed according to an association relationship between the plurality of first user nodes;

[0022] The first attention obtaining module is configured to process the first graph structure features and the first user features to obtain an attention parameter corresponding to each first user node in the first target graph, where the attention parameter is used to represent an importance of the first user node in the first target graph.

[0023] The first screening module is configured to select a plurality of second user nodes from the plurality of first user nodes according to the obtained attention parameters, where the attention parameters corresponding to the plurality of second user nodes are greater than the attention parameters corresponding to other first user nodes.

[0024] The type identifying module is configured to identify the first user features of the plurality of second user nodes and second graph structure features of a second target graph to obtain a group type of the target group, where the second target graph is constructed according to an association relationship between the plurality of second user nodes.

[0025] In a possible implementation manner, the apparatus further includes:

[0026] The feature adjusting module is configured to adjust the first user features of the plurality of second user nodes according to the second graph structure features to obtain second user features of the plurality of second user nodes.

[0027] The second attention obtaining module is configured to process the second graph structure features and the second user features to obtain an attention parameter of each second user node in the second target graph.

[0028] The second screening module is configured to select a plurality of third user nodes from the plurality of second user nodes according to the obtained attention parameters, where the attention parameters corresponding to the plurality of third user nodes are greater than the attention parameters corresponding to other second user nodes.

[0029] In another possible implementation manner, the type identifying module is configured to identify the first user features of the plurality of second user nodes, the second graph structure features, second user features of the plurality of third user nodes, and third graph structure features of a third target graph to obtain a group type of the target group, where the third target graph is constructed according to an association relationship between the plurality of third user nodes.

[0030] In another possible implementation manner, the type identifying module includes:

[0031] The first fusion unit is configured to fuse the first user features of the plurality of second user nodes and the second graph structure features to obtain first fused features.

[0032] a second fusion unit, configured to fuse the second user features of the plurality of third user nodes and the third graph structure feature to obtain second fusion features;

[0033] a type identification unit, configured to identify the first fusion features and the second fusion features to obtain a group type of the target group.

[0034] In another possible implementation, the first fusion unit is configured to:

[0035] perform average processing on the first user features of the plurality of second user nodes and the second graph structure feature according to a quantity of the second user nodes to obtain average user features corresponding to the plurality of second user nodes;

[0036] splice the average user features and maximum user features in the first user features of the plurality of second user nodes to obtain the first fusion features.

[0037] In another possible implementation, the type identification unit is configured to:

[0038] splice the first fusion features and the second fusion features to obtain spliced features corresponding to the target group;

[0039] identify the spliced features to obtain the group type of the target group.

[0040] In another possible implementation, the group type identification model includes a first attention network, a first screening network, and an identification network,

[0041] The first attention obtaining module is configured to invoke the first attention network to process the first graph structure feature and the plurality of first user features to obtain attention parameters corresponding to each first user node in the first target graph.

[0042] The first screening module is configured to invoke the first screening network to select the plurality of second user nodes from the plurality of first user nodes according to the obtained plurality of attention parameters.

[0043] The type identification module is configured to invoke the identification network to identify the first user features of the plurality of second user nodes and the second graph structure feature to obtain the group type of the target group.

[0044] In another possible implementation, the group type identification model further includes a first convolution network, a second attention network, and a second screening network, and the apparatus further includes:

[0045] The feature adjustment module is configured to call the first convolutional network, and adjust first user features of the plurality of second user nodes according to the second graph structure features to obtain second user features of the plurality of second user nodes.

[0046] The second attention obtaining module is configured to call the second attention network, and process the second graph structure features and the plurality of second user features to obtain attention parameters of each second user node in the second target graph.

[0047] The second screening module is configured to call the second screening network, and select a plurality of third user nodes from the plurality of second user nodes according to the obtained plurality of attention parameters, the plurality of third user nodes corresponding to attention parameters greater than attention parameters of other second user nodes.

[0048] In another possible implementation, the type identification module is configured to call the identification network, and identify the first user features of the plurality of second user nodes, the second graph structure features, the second user features of the plurality of third user nodes, and third graph structure features of a third target graph to obtain a group type of the target group, the third target graph being constructed according to an association relationship between the plurality of third user nodes.

[0049] In another possible implementation, the group type identification model further includes a first fusion network and a second fusion network, and the type identification module includes:

[0050] The first fusion unit is configured to call the first fusion network, and fuse the first user features of the plurality of second user nodes and the second graph structure features to obtain first fusion features.

[0051] The second fusion unit is configured to call the second fusion network, and fuse the second user features of the plurality of third user nodes and the third graph structure features to obtain second fusion features.

[0052] The type identification unit is configured to call the identification network, and identify the first fusion features and the second fusion features to obtain the group type of the target group.

[0053] In another possible implementation, the group type identification model further includes a splicing network, and the type identification unit is configured to:

[0054] call the splicing network to splice the first fusion features and the second fusion features to obtain splicing features corresponding to the target group;

[0055] call the identification network to identify the splicing features to obtain the group type of the target group.

[0056] In another possible implementation, the training process of the group type identification model comprises:

[0057] obtaining a sample type of a sample group, a sample graph structure feature of a sample graph, and sample user features of a plurality of sample user nodes in the sample target graph, the sample user nodes being nodes corresponding to the sample user identifiers, and the sample graph being constructed according to association relationships between the plurality of sample user identifiers in the sample group;

[0058] calling the group type identification model to identify the sample graph structure feature and the sample user features of the plurality of sample user nodes, to obtain a predicted type of the sample group;

[0059] training the group type identification model according to a difference between the sample type and the predicted type.

[0060] In another possible implementation, the first screening network is configured to:

[0061] multiply the number of the plurality of first user nodes by a reference ratio to obtain a reference number;

[0062] arrange the attention parameters of the plurality of first user nodes in descending order, select a reference number of attention parameters arranged in front, and select the first user nodes corresponding to the selected attention parameters as the plurality of second user nodes.

[0063] In another possible implementation, the graph structure feature comprises an association degree feature between any two user nodes in the plurality of user nodes, and the feature acquisition module is configured to:

[0064] obtain a common occurrence frequency of any two user identifiers in the target group, the common occurrence frequency being a frequency of the any two user identifiers publishing content in the target group together within a plurality of reference time periods;

[0065] determine an association degree feature between the any two user identifiers according to the common occurrence frequency, the association degree feature being in a positive correlation relationship with the common occurrence frequency.

[0066] In another possible implementation, the user feature comprises a user behavior feature and a user attribute feature, and the feature acquisition module is configured to:

[0067] obtain a user social network, the user social network comprising a plurality of registered user identifiers;

[0068] obtain a user behavior feature of the plurality of user identifiers in the target group according to the user social network;

[0069] According to the user portrait information corresponding to the plurality of user identifiers in the target group, user attribute features of the plurality of user identifiers in the target group are obtained.

[0070] In another aspect, a computer device is provided, which includes a processor and a memory, the memory storing at least one computer program, the at least one computer program being loaded and executed by the processor to implement the operations performed in the group type identification method according to the above aspects.

[0071] In another aspect, a computer readable storage medium is provided, which stores at least one computer program, the at least one computer program being loaded and executed by a processor to implement the operations performed in the group type identification method according to the above aspects.

[0072] In another aspect, a computer program product or a computer program is provided, which includes computer program code stored in a computer readable storage medium, the computer program code being read by a processor of a computer device from the computer readable storage medium, and the processor executes the computer program code to cause the computer device to implement the operations performed in the group type identification method according to the above aspects.

[0073] The technical scheme provided by the embodiments of the present application has at least the following beneficial effects:

[0074] The method provided by the embodiments of the present application considers the graph structure features and the user node features to obtain the attention parameters of each first user node, compared with only obtaining user information in the related art, the amount of information is increased, so that the obtained attention parameters can more accurately reflect the importance of the user nodes in the graph structure, so that when the plurality of first user nodes are screened according to the reference attention parameters, the important user nodes can be accurately selected, the target group is identified according to the user features and the graph structure features of these important user nodes, to improve the identification accuracy, and discarding the unimportant user nodes can reduce the amount of data processed and improve the processing speed. BRIEF DESCRIPTION OF DRAWINGS

[0075] In order to more clearly illustrate the technical scheme in the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced as follows. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0076] Figure 1is a flowchart of a group type identification method provided by an embodiment of the present application;

[0077] Figure 2 is a structural schematic diagram of a group type identification model provided by an embodiment of the present application;

[0078] Figure 3 is a structural schematic diagram of another group type identification model provided by an embodiment of the present application;

[0079] Figure 4 is a structural schematic diagram of another group type identification model provided by an embodiment of the present application;

[0080] Figure 5 is a flowchart of another group type identification method provided by an embodiment of the present application;

[0081] Figure 6 is a structural schematic diagram of another group type identification model provided by an embodiment of the present application;

[0082] Figure 7 is a structural schematic diagram of a group type identification device provided by an embodiment of the present application;

[0083] Figure 8 is a structural schematic diagram of another group type identification device provided by an embodiment of the present application;

[0084] Figure 9 is a structural schematic diagram of a terminal provided by an embodiment of the present application;

[0085] Figure 10 is a structural schematic diagram of a server provided by an embodiment of the present application. DETAILED DESCRIPTION

[0086] In order to make the objectives, technical solutions and advantages of embodiments of the present application clearer, the following will further describe the embodiments of the present application with reference to the accompanying drawings.

[0087] It can be understood that the terms "first", "second", etc. used in the present application can be used herein to describe various concepts, but unless specifically stated, these concepts are not limited by these terms. These terms are only used to distinguish one concept from another. For example, without departing from the scope of the present application, a first user node can be referred to as a second user node, and a second user node can be referred to as a first user node.

[0088] As used herein, the terms “at least one”, “multiple”, “each”, “any”, and the like, include one, two, or more than two, multiple includes two or more than two, each refers to each of the corresponding multiple, and any refers to any of the multiple. For example, a plurality of user nodes includes 3 user nodes, each user node refers to each of the 3 user nodes, and any refers to any of the 3 user nodes, which can be the first, the second, or the third.

[0089] Artificial Intelligence (AI) is the theory, method, technology and application system of using digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive environment, acquire knowledge and use knowledge to obtain the best results. In other words, artificial intelligence is a comprehensive technology of computer science, which attempts to understand the essence of intelligence and produce a new intelligent machine that can react in a similar way to human intelligence. Artificial intelligence is to study the design principles and implementation methods of various intelligent machines, so that machines have the functions of perception, reasoning and decision-making.

[0090] Artificial intelligence technology is a comprehensive discipline, involving a wide range of fields, both hardware and software technologies. Artificial intelligence basic technologies generally include technologies such as sensors, special artificial intelligence chips, cloud computing, distributed storage, big data processing technology, operation / interaction system, mechatronics, etc. Artificial intelligence software technology mainly includes computer vision technology, speech processing technology, natural language processing technology, and machine learning / deep learning, etc.

[0091] Nature Language processing (NLP) is an important direction in the field of computer science and artificial intelligence. It studies various theories and methods that can realize effective communication between people and computers using natural language. Natural language processing is a science that integrates linguistics, computer science and mathematics. Therefore, the research in this field will involve natural language, i.e. the language used in daily life, so it has a close relationship with the study of linguistics. Natural language processing technology usually includes text processing, semantic understanding, machine translation, robot question and answer, knowledge graph, etc.

[0092] Machine Learning (ML) is a multi-disciplinary subject that involves probability theory, statistics, approximation theory, convex analysis, algorithmic complexity theory, etc. It is a specialized study of how computers simulate or implement human learning behavior to acquire new knowledge or skills, reorganize existing knowledge structure, and continuously improve their performance. Machine learning is the core of artificial intelligence and the fundamental approach to making computers intelligent. It is applied in various fields of artificial intelligence. Machine learning and deep learning usually include artificial neural networks, belief networks, reinforcement learning, transfer learning, inductive learning, and teaching learning.

[0093] With the research and progress of artificial intelligence technology, artificial intelligence technology is researched and applied in many fields, such as common smart home, smart wearable device, virtual assistant, smart speaker, smart marketing, unmanned vehicle, autonomous vehicle, unmanned aerial vehicle, robot, smart medical treatment, smart customer service, etc. It is believed that with the development of technology, artificial intelligence technology will be applied in more fields and play an increasingly important role.

[0094] The scheme provided by the embodiments of the present application relates to machine learning and other technologies of artificial intelligence. The embodiments are specifically described as follows.

[0095] The group type identification method provided by the embodiments of the present application can be applied in a group type identification scenario. For example, in an instant messaging application, in order to protect user privacy and user property safety, the group type identification method provided by the embodiments of the present application is used to identify whether a target group is a sensitive group. If it is determined that the target group is a sensitive group, the target group is processed by group sealing or other processing to avoid loss to the user.

[0096] The method provided by the embodiments of the present application is applied to a computer device, which is a terminal or a server. Optionally, the terminal is a portable, pocket-sized, handheld, or other type of terminal, such as a mobile phone, a computer, a tablet computer, etc. The server is a stand-alone physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN (Content Delivery Network), and basic cloud computing services such as big data and artificial intelligence platforms.

[0097] Figure 1 is a flowchart of a group type identification method provided by the embodiments of the present application. The execution subject of the embodiments of the present application is a computer device. Referring to Figure 1 , the method comprises the following steps:

[0098] 101. Obtain a first graph structural feature of a first target graph and first user features of multiple first user nodes in the first target graph.

[0099] In the embodiment of the present application, a target group including multiple user identifiers is taken as an example for description.

[0100] The first user node is a node corresponding to a user identifier in the target group, and the first target graph is constructed based on the association relationship between multiple first user nodes.

[0101] In one possible implementation, if the multiple first user nodes are user nodes corresponding to all user identifiers in the target group, that is, the multiple first user nodes are not screened, then the first target graph is the initial target graph; if the multiple first user nodes are user nodes corresponding to some user identifiers in the target group, that is, the multiple first user nodes have been obtained after screening, then the first target graph is the target graph obtained after screening.

[0102] The first graph structure feature is used to describe the first target graph, which includes multiple first user nodes and connecting lines between the multiple first user nodes. A connecting line between any two first user nodes indicates that there is an association relationship between the two first user nodes. The first user feature is used to describe the user corresponding to the user identifier, and the first user feature includes at least one of a user behavior feature or a user attribute feature.

[0103] 102. Process the first graph structure feature and the multiple first user features to obtain an attention parameter corresponding to each first user node in the first target graph.

[0104] The attention parameter is used to represent the importance of the first user node in the first target graph, that is, the importance of the user ID corresponding to the first user node in the target group. The importance of the user ID in the target group is positively correlated with the attention parameter. For example, for users such as group owners and administrators in the target group, they are more important in the target group, and the attention parameters of the user nodes corresponding to these users will also be larger.

[0105] 103. Select multiple second user nodes from the multiple first user nodes based on the obtained multiple attention parameters.

[0106] Among them, the attention parameters corresponding to multiple second user nodes are greater than the attention parameters corresponding to other first user nodes.

[0107] In the embodiments of the present application, the plurality of first user nodes are screened, and a plurality of second user nodes with larger attention parameters are selected from the plurality of first user nodes. Subsequently, the plurality of second user nodes with larger attention parameters are processed, so as to discard some unimportant information in the processing process, thereby ensuring that the data amount is reduced while the loss of important information is reduced.

[0108] 104. The first user features of the plurality of second user nodes and the second graph structure features of the second target graph are identified to obtain a group type of the target group.

[0109] The second target graph is constructed according to the association relationship between the plurality of second user nodes, and the second graph structure features are used to describe the second target graph. The second target graph is a subgraph of the first target graph, and the second target graph includes the plurality of second user nodes and the connection lines between the plurality of second user nodes. The connection lines between the plurality of second user nodes in the second target graph are the same as the connection lines of the plurality of second user nodes in the first target graph.

[0110] The group type refers to the type to which the target group belongs. Different group types can be obtained according to different division standards. For example, according to the occupation of the user corresponding to the user identifier in the target group, the work group and the non-work group can be divided; according to whether the users corresponding to the user identifiers in the target group are relatives, the family group and the non-family group can be divided; according to whether the behavior of the user corresponding to the user identifier in the target group is a sensitive behavior, the sensitive group and the non-sensitive group can be divided. For example, the sensitive behavior includes behaviors related to pornography, gambling, fraud, etc.

[0111] The method provided in the embodiments of the present application considers the graph structure features and the user node features to obtain the attention parameters of each first user node. Compared with the related art which only obtains user information, the amount of information is increased, so that the obtained attention parameters can more accurately reflect the importance of the user node in the graph structure. Therefore, when the plurality of first user nodes are screened according to the reference attention parameters, the more important user nodes can be accurately selected, the target group is identified according to the user features and the graph structure features of the important user nodes, so as to improve the identification accuracy. Meanwhile, discarding the unimportant user nodes can reduce the data amount for processing and improve the processing speed.

[0112] In a possible implementation manner, the computer device calls the group type identification model to identify the group type of the target group. The model structure of the group type identification model is described first as follows.

[0113] Figure 2 is a structural schematic diagram of a group type identification model provided in the embodiments of the present application. Referring to Figure 2The group type identification model includes an input network 201, a first attention network 202, a first screening network 203, and an identification network 204. The input network 201 is connected to the first attention network 202, the first attention network 202 is connected to the first screening network 203, and the first screening network 203 is connected to the identification network 204.

[0114] The input network 201 is configured to obtain input graph structure features and user features, the first attention network 202 is configured to obtain attention parameters of each user node, the first screening network 203 is configured to screen important user nodes according to the obtained attention parameters, and the identification network 204 is configured to identify group types according to user features of the screened user nodes and corresponding graph structure features.

[0115] In a possible implementation, referring to Figure 3 The group type identification model further includes a first convolution network 205, a second attention network 206, a second screening network 207, and a splicing network 208. The first convolution network 205 is connected to the first screening network 203 and the second attention network 206, the second attention network 206 is connected to the second screening network 207, the second screening network 207 is connected to the identification network 204, and the splicing network 208 is connected to the first screening network 203, the second screening network 207, and the identification network 204.

[0116] The first convolution network 205 is configured to further process user features of comparatively important user nodes screened by the first screening network 203, the second attention network 206 is configured to obtain attention parameters of each screened user node, the second screening network 207 is configured to further screen the screened user nodes according to the obtained attention parameters, and the splicing network 208 is configured to splice user features of the user nodes screened by the first screening network 203 and user features of the user nodes screened again by the second screening network 207.

[0117] The above possible implementation is described by taking twice screening of user nodes as an example. In another possible implementation, the user nodes can be screened three times, four times, or even more times, so as to obtain more information quantity. For example, referring to Figure 4 Three attention networks and three screening networks are used to screen the user nodes three times. The splicing network is configured to splice user features of the user nodes screened by the three screening networks.

[0118] In addition, in a possible implementation, a convolution network can also be arranged before the first attention network, that is, the input user features and graph structure features are first subjected to convolution processing, and the first attention network obtains the attention parameters according to the convolution-processed user features and graph structure features.

[0119] Below through Figure 5 The illustrated embodiment describes calling a group type identification model to identify the type of a target group.

[0120] Figure 5 This is a flow chart of a group type identification method provided by an embodiment of the present application. The execution subject of the embodiment of the present application is a computer device. Figure 5 , the method comprises the following steps:

[0121] 501. Construct a first target graph according to association relationships between multiple user identifiers in a target group.

[0122] The embodiment of the present application is described by taking the first target graph as an example of the initially constructed target graph.

[0123] In an embodiment of the present application, corresponding user nodes are constructed based on multiple user identifiers in the target group, and then the multiple user nodes are connected together based on the association relationship between the multiple user identifiers to form a first target graph. The first target graph is an isomorphic graph, that is, the first target graph only includes nodes of one type, namely user nodes, and the user nodes are connected based on the same type of association relationship, that is, the connecting lines in the first target graph are connecting lines of the same type.

[0124] The association relationship between any two user nodes is determined based on a correlation feature between the two user nodes, and the correlation feature is used to represent the degree of intimacy between users corresponding to the two user nodes.

[0125] In one possible implementation, the number of co-occurrences of any two user identifiers in the target group is obtained, and based on the number of co-occurrences, a correlation feature between the two user identifiers is determined. The number of co-occurrences refers to the number of times that any two user identifiers jointly publish content in the target group within multiple reference time periods. The correlation feature is positively correlated with the number of co-occurrences: that is, the greater the number of co-occurrences, the greater the correlation between the two user identifiers, and the smaller the number of co-occurrences, the smaller the correlation between the two user identifiers. Optionally, the number of co-occurrences can be directly used as the correlation feature.

[0126] For example, if the reference time period is 5 minutes, for any two users, determine whether they both spoke in the target group within 5 minutes. If both users spoke, the co-occurrence count for these two users is increased by 1. If only one user spoke, or neither user spoke, the co-occurrence count does not increase. If two users both speak in the target group, it indicates that they communicate frequently and are close friends. Therefore, the co-occurrence count can be used to determine the correlation feature.

[0127] In a possible implementation, it is considered that there is a possibility that two users speak at the same time in the target group due to a coincidence, and therefore it is further needed to determine whether the co-occurrence number is less than a reference number, and in response to the co-occurrence number being less than the reference number, the co-occurrence number is set to 0, so as to avoid the coincidence. For example, a group owner in the target group issues a message, and other users in the target group may all reply to the message, and at this time, many users in the target group may speak at the same time, but the users speaking at the same time may not be close to each other.

[0128] It should be noted that, in the embodiment of the present application, the first target graph is taken as an example of a homogeneous graph, and in another embodiment, the first target graph can also be a heterogeneous graph. For example, the first target graph includes user nodes and user type nodes, the user nodes are connected according to the association relationship of the same type, and the user nodes and the user type nodes are connected according to the type to which the user nodes correspond.

[0129] 502、obtaining a first graph structure feature of the first target graph and a first user feature of a plurality of first user nodes in the first target graph.

[0130] The first graph structure feature is used to represent a plurality of user nodes in the first target graph and an association relationship between the plurality of user nodes, and the first graph structure feature includes an association degree feature between any two user nodes in the plurality of user nodes. The first user feature includes a user behavior feature and a user attribute feature, the user behavior feature is used to represent a historical behavior of a user, and the user attribute feature is used to represent an attribute of the user itself.

[0131] In a possible implementation, a user social network is obtained, and according to the user social network, a user behavior feature of a plurality of user identifiers in the target group is obtained. The user social network includes a plurality of registered user identifiers.

[0132] Optionally, a graph embedding method is used to obtain the user behavior feature of the user identifier according to the user social network. The graph embedding is used to represent each node in the graph as a dense vector in a low-dimensional space, and the obtained dense vector is used as the feature information of the node. The core idea of the graph embedding is to retain the internal structural properties of the graph structure, that is, to keep the nodes connected in the graph close to each other in the vector space. For example, the graph embedding method includes a DeepWalk method (a method for generating node representations in a network) and a Node2Vec method (a model for generating node vectors in a network), and the like.

[0133] Taking the Deepwalk node embedding method as an example, starting from each user node in the graph, according to the user social network and the user connection weight, multiple trajectories are randomly walked. All the trajectories are input into the word2vec word vector embedding model as a corpus, and processed by the word2vec word vector embedding model to finally obtain the user behavior characteristics of each user node.

[0134] In one possible implementation, user attribute features of multiple user identifiers in the target group are obtained based on user profile information corresponding to the multiple user identifiers in the target group. The user profile information includes user avatars, dynamic information posted by users, user age, gender, and user geographic location.

[0135] In one possible implementation, a vector is used to represent the first user feature, and a matrix is ​​used to represent multiple first user features. That is, multiple first user features are concatenated together to form a matrix. For example, a row or a column in the matrix represents the first user feature of a first user node.

[0136] In one possible implementation, a matrix is ​​used to represent the first graph structure feature, and each element in the matrix represents the association relationship between the corresponding two user nodes. For example, the element in the third row and fourth column of the matrix represents the association relationship between the third user node and the fourth user node. For example, the following formula is used to obtain the association relationship between any two first user nodes:

[0137] A ij =log(C ij );

[0138] Among them, A ij represents the association relationship between the i-th first user node and the j-th user node, C ij The number of times the user identifier corresponding to the jth first user node and the user identifier corresponding to the jth user node co-occur.

[0139] 503. Call the first attention network to process the first graph structure feature and the multiple first user features to obtain an attention parameter corresponding to each first user node in the first target graph.

[0140] The first attention network is at least one GNN (Graph Neural Network). Use any of the following formulas to obtain the attention parameters:

[0141] Z = σ(GNN(X,A));

[0142] Z=σ(GNN(X,A+A 2 ));

[0143] Z=σ(GNN2(σ(GNN1(X,A)),A));

[0144]

[0145] Where Z represents the attention parameter, X represents multiple first user features, A represents the first graph structure feature, σ is a reference value, GNN(·) represents the convolution of the first user feature and the first graph structure feature, and m represents the mth GNN. Z, X, and A are matrices.

[0146] The first and second formulas above use a single GNN, the third formula uses two GNNs, and the fourth formula uses M GNNs. In the third formula, the two GNNs are connected sequentially, meaning the first GNN performs a single pass, followed by a second pass based on the first pass, to obtain the attention parameter. In the fourth formula, the M GNNs process the first user feature and the first graph structure feature, respectively. The resulting M attention parameters are averaged to obtain the final attention parameter.

[0147] 504. Call the first screening network, select multiple second user nodes from the multiple first user nodes according to the obtained multiple attention parameters, and construct a second target graph according to the association relationship between the multiple second user nodes.

[0148] Among them, the attention parameters corresponding to multiple second user nodes are greater than the attention parameters corresponding to other first user nodes.

[0149] In one possible implementation, the first screening network is invoked to multiply the number of first user nodes by a reference ratio to obtain a reference number. The attention parameters of the first user nodes are arranged in descending order, and the first user nodes corresponding to the selected attention parameters are selected as the selected second user nodes. After the selected second user nodes are selected, a second target graph is constructed based on the associations among the selected second user nodes.

[0150] Optionally, the following formula is used to obtain the first user features of the plurality of second user nodes and the second graph structure features of the second target graph:

[0151] X′=X idx ,;A l =A idx,idx ;

[0152] Where X′ represents the first user features of multiple second user nodes, X idxrepresents selecting the first user feature of the second user node from the first user features of the plurality of first user nodes, A l represents the second graph structure feature, A idx，idx represents selecting the association relationship corresponding to the plurality of second user nodes from the first graph structure feature.

[0153] Optionally, considering the size of the attention parameter, the first user features of the plurality of second user nodes are adjusted by using the following formula to obtain adjusted first user features:

[0154] X l = X'☉Z;

[0155] wherein, X l represents the adjusted first user feature, Z represents the attention parameter of the plurality of second user nodes, and ☉ represents bit multiplication, that is, multiplying the first user feature of each second user node by the corresponding attention parameter.

[0156] It should be noted that the present embodiment only takes the multiple screening of the user nodes as an example for description, and in another embodiment, if the user nodes are screened only once, after the plurality of second user nodes are selected, the group type of the target group is obtained according to the user features of the plurality of second user nodes and the second graph structure feature of the second target graph, and the subsequent steps are no longer executed.

[0157] 505, calling a first convolutional network, adjusting the first user features of the plurality of second user nodes according to the second graph structure feature to obtain second user features of the plurality of second user nodes.

[0158] The first convolutional network is a GCN (Graph Convolutional Networks, graph convolutional network), and the following formula is used to determine the second user features after convolution processing:

[0159] X l+1 = σ(A l X l W l+1 );

[0160] wherein, X l+1 represents the second user features of the plurality of second user nodes, X l represents the first user features of the plurality of second user nodes, A l represents the second graph structure feature, and W l+1 represents the model parameters in the first convolutional network.

[0161] 506, calling a second attention network, processing the second graph structure feature and the plurality of second user features to obtain the attention parameter of each second user node in the second target graph.

[0162] 507. calling the second screening network to select a plurality of third user nodes from the plurality of second user nodes according to the obtained plurality of attention parameters, and constructing a third target graph according to the association relationships among the plurality of third user nodes.

[0163] The implementation of steps 506-507 is the same as that of steps 503-504, and thus is not described herein again.

[0164] 508. calling the recognition network to recognize the first user features of the plurality of second user nodes, the second graph structure features, the second user features of the plurality of third user nodes, and the third graph structure features of the third target graph, to obtain the group type of the target group.

[0165] In a possible implementation, to reduce the amount of data processed by the recognition network, the first user features and the second graph structure features of the plurality of second user nodes are fused to obtain first fused features, the second user features and the third graph structure features of the plurality of third user nodes are fused to obtain second fused features, and the recognition network is called to recognize the first fused features and the second fused features to obtain the group type of the target group.

[0166] In a possible implementation, according to the number of second user nodes, the first user features and the second graph structure features of the plurality of second user nodes are averaged to obtain average user features corresponding to the plurality of second user nodes, and the average user features are spliced with maximum user features in the first user features of the plurality of second user nodes to obtain the first fused features.

[0167] For example, the first fused features are obtained by using the following formula:

[0168]

[0169] wherein s represents the first fused features, N represents the number of second user nodes, x i represents the first user features of the i th second user node and the association relationships with other second user nodes, and || represents splicing the preceding features and the following features.

[0170] Similarly, for the second user features and the third graph structure features of the third user nodes, according to the number of third user nodes, the second user features and the third graph structure features of the plurality of third user nodes are averaged to obtain average user features corresponding to the plurality of third user nodes, and the average user features are spliced with maximum user features in the second user features of the plurality of third user nodes to obtain the second fused features.

[0171] In a possible implementation, the first fusion feature and the second fusion feature are spliced to obtain a spliced feature corresponding to the target group, and the spliced feature is identified to obtain a group type corresponding to the target group. The spliced feature is a feature representing the target group.

[0172] In a possible implementation, referring to the model structure diagram shown in FIG. 2, the group type identification model further includes a first fusion network, a second fusion network, and a splicing network, that is, the first fusion network is invoked to fuse the first user features and the second graph structure features of the plurality of second user nodes to obtain first fusion features; the second fusion network is invoked to fuse the second user features and the third graph structure features of the plurality of third user nodes to obtain second fusion features; the splicing network is invoked to splice the first fusion features and the second fusion features to obtain a spliced feature corresponding to the target group; and the identification network is invoked to identify the spliced feature to obtain a group type of the target group. Figure 6

[0173] Taking determining whether the target group is a sensitive group as an example, in a possible implementation, the identification network can be a classifier, and the classifier includes a multi-layer perceptron. The output of the identification network is 0 or 1. If the output of the identification network is 0, it indicates that the target group is not a sensitive group. If the output of the identification network is 1, it indicates that the target group is a sensitive group. For example, the output value of the identification network is determined by using the following formula:

[0174]

[0175] wherein, The output value of the identification network is represented by y, E represents the spliced feature, and MLP(·) represents that the spliced feature is processed by using an MLP (Multi-Layer Perceptron).

[0176] In another possible implementation, the output of the identification network is a probability. If the output probability is greater than a reference probability, it indicates that the target group is a sensitive group. If the output probability is not greater than the reference probability, it indicates that the target group is not a sensitive group.

[0177] ​The method provided in the embodiments of the present application considers the graph structure features and the user node features to obtain the attention parameter of each first user node, and compared with only obtaining user information in the related art, the amount of information is increased, so that the obtained attention parameter can more accurately reflect the importance of the user node in the graph structure, thereby when the plurality of first user nodes are screened according to the reference attention parameter, the relatively important user nodes can be accurately selected, the target group is identified according to the user features and the graph structure features of the important user nodes, so as to improve the identification accuracy, and meanwhile, discarding the unimportant user nodes can reduce the amount of data processed and improve the processing speed.

[0178] In addition, in the embodiments of the present application, the user nodes are screened multiple times to obtain user features and graph structure features of different levels, and when the target group is identified, the user features and the graph structure features of different levels are considered, so that the identification accuracy is further improved.

[0179] In a possible implementation manner, before the group type identification model is used, the group type identification model needs to be trained first, and the training process of the group type identification model includes: obtaining a sample type of a sample group, a sample graph structure feature of a sample graph, and sample user features of a plurality of sample user nodes in a sample target graph; calling the group type identification model to identify the sample graph structure feature and the sample user features of the plurality of sample user nodes to obtain a predicted type of the sample group; and training the group type identification model according to a difference between the sample type and the predicted type. The sample user node is a node corresponding to a sample user identifier, and the sample graph is constructed according to an association relationship between the plurality of sample user identifiers in the sample group.

[0180] For example, the square error is used to train the group type identification model.

[0181]

[0182] wherein, represents the predicted type, Y represents the sample type, L i represents the difference between the predicted type and the sample type.

[0183] Optionally, after the group type of the target group is identified by calling the group type identification model, the target group can be used to continue training the group type identification model.

[0184] In a possible implementation, for the application scenario of determining whether the target group is a sensitive group, in the prior art, a keyword filtering technology is used to determine whether the content published by the user includes a sensitive keyword, and if the sensitive keyword is included, it is considered that the target group belongs to the sensitive group, but if the user uses other non-sensitive keywords instead of the corresponding sensitive keywords, the keyword filtering technology cannot detect it, and it cannot be determined whether it is a sensitive keyword, and the identification accuracy of the target group is low. In another related technology, a user reporting method is used, that is, after the user reports the target group, a technical personnel manually reviews and determines whether the target group belongs to the sensitive group, but this method depends on the user's reporting and is affected by the user's reporting behavior, and the identification accuracy of the target group is also low. The method provided in the embodiments of the present application is compared with the above two related technologies, is not affected by the words used by the user and the user reporting behavior, and can directly identify the target group according to the association relationship between the multiple user nodes in the target group and the user features of the user nodes, thereby improving the identification accuracy.

[0185] In another related technology, a pooling function is used to read the information of all user nodes in the target group, but when the user nodes are many, it is difficult to read all the information by using the pooling function, which may cause a large amount of information to be lost, and the more user nodes, the slower the processing speed. In the present application, the user nodes can be screened according to the attention parameters, and only the information of the screened user nodes needs to be read out, and the information of important nodes can be ensured to be retained according to the size of the attention parameters, so that a large amount of information is not lost, and the processing speed is improved while the amount of information is ensured.

[0186] Figure 7 is a structural schematic diagram of a group type identification device provided by the embodiments of the present application. Referring to Figure 7 The device comprises:

[0187] The feature acquisition module 701 is configured to acquire a first graph structure feature of a first target graph and first user features of multiple first user nodes in the first target graph, the first user nodes being nodes corresponding to user identifiers in the target group, and the first target graph being constructed according to association relationships between the multiple first user nodes.

[0188] The first attention acquisition module 702 is configured to process the first graph structure feature and the multiple first user features to obtain an attention parameter corresponding to each first user node in the first target graph, the attention parameter being used to represent an importance of the first user node in the first target graph.

[0189] The first screening module 703 is configured to select, according to the obtained plurality of attention parameters, a plurality of second user nodes from the plurality of first user nodes, the plurality of second user nodes corresponding to attention parameters greater than attention parameters corresponding to other first user nodes.

[0190] The type identification module 704 is configured to identify the first user features of the plurality of second user nodes and the second graph structure features of the second target graph to obtain a group type of the target group, the second target graph being constructed according to the association relationships between the plurality of second user nodes.

[0191] The device provided in the embodiments of the present application considers the graph structure features and the user node features to obtain the attention parameters of each first user node, and compared with the related art that only obtains user information, the amount of information is increased, so that the obtained attention parameters can more accurately reflect the importance of the user nodes in the graph structure, thereby when the plurality of first user nodes are screened according to the reference attention parameters, the more important user nodes can be accurately selected, and the target group is identified according to the information of the important user nodes, so as to improve the identification accuracy, and meanwhile, discarding the unimportant user nodes can reduce the amount of data processed and improve the processing speed.

[0192] In a possible implementation manner, referring to Figure 8 The device further includes:

[0193] The feature adjustment module 705 is configured to adjust the first user features of the plurality of second user nodes according to the second graph structure features to obtain second user features of the plurality of second user nodes.

[0194] The second attention obtaining module 706 is configured to process the second graph structure features and the plurality of second user features to obtain attention parameters of each second user node in the second target graph.

[0195] The second screening module 707 is configured to select, according to the obtained plurality of attention parameters, a plurality of third user nodes from the plurality of second user nodes, the plurality of third user nodes corresponding to attention parameters greater than attention parameters corresponding to other second user nodes.

[0196] In another possible implementation manner, the type identification module 704 is configured to identify the first user features of the plurality of second user nodes, the second graph structure features, the second user features of the plurality of third user nodes, and third graph structure features of a third target graph to obtain a group type of the target group, the third target graph being constructed according to association relationships between the plurality of third user nodes.

[0197] In another possible implementation manner, referring to Figure 8 The type identification module 704 includes:

[0198] The first fusion unit 7041 is configured to fuse the first user features and the second graph structure features of the plurality of second user nodes to obtain first fusion features.

[0199] The second fusion unit 7042 is configured to fuse the second user features and the third graph structure features of the plurality of third user nodes to obtain second fusion features.

[0200] The type identification unit 7043 is configured to identify the first fusion features and the second fusion features to obtain the group type of the target group.

[0201] In another possible implementation, referring to Figure 8 The first fusion unit 7041 is configured to:

[0202] According to the number of the second user nodes, the first user features and the second graph structure features of the plurality of second user nodes are averaged to obtain average user features corresponding to the plurality of second user nodes;

[0203] The average user features are spliced with maximum user features in the first user features of the plurality of second user nodes to obtain the first fusion features.

[0204] In another possible implementation, referring to Figure 8 The type identification unit 7043 is configured to:

[0205] The first fusion features and the second fusion features are spliced to obtain spliced features corresponding to the target group;

[0206] The spliced features are identified to obtain the group type of the target group.

[0207] In another possible implementation, referring to Figure 8 The group type identification model includes a first attention network, a first screening network, and an identification network,

[0208] The first attention obtaining module 702 is configured to invoke the first attention network to process the first graph structure features and the plurality of first user features to obtain attention parameters corresponding to each first user node in the first target graph;

[0209] The first screening module 703 is configured to invoke the first screening network to select the plurality of second user nodes from the plurality of first user nodes according to the plurality of obtained attention parameters;

[0210] The type identification module 704 is configured to invoke the identification network to identify the first user features and the second graph structure features of the plurality of second user nodes to obtain the group type of the target group.

[0211] In another possible implementation manner, the group type identification model further includes a first convolutional network, a second attention network, and a second screening network, referring to Figure 8 The apparatus further includes:

[0212] The feature adjustment module 705 is configured to invoke the first convolutional network, and adjust the first user features of the plurality of second user nodes according to the second graph structure features to obtain second user features of the plurality of second user nodes.

[0213] The second attention acquisition module 706 is configured to invoke the second attention network, and process the second graph structure features and the plurality of second user features to obtain an attention parameter of each second user node in the second target graph.

[0214] The second screening module 707 is configured to invoke the second screening network, and select a plurality of third user nodes from the plurality of second user nodes according to the plurality of obtained attention parameters, the attention parameters of the plurality of third user nodes being greater than the attention parameters of other second user nodes.

[0215] In another possible implementation manner, the type identification module 704 is configured to invoke the identification network, and identify the first user features of the plurality of second user nodes, the second graph structure features, the second user features of the plurality of third user nodes, and third graph structure features of a third target graph to obtain the group type of the target group, the third target graph being constructed according to the association relationships among the plurality of third user nodes.

[0216] In another possible implementation manner, the group type identification model further includes a first fusion network and a second fusion network, referring to Figure 8 The type identification module 704 includes:

[0217] The first fusion unit 7041 is configured to invoke the first fusion network, and fuse the first user features of the plurality of second user nodes and the second graph structure features to obtain first fusion features.

[0218] The second fusion unit 7042 is configured to invoke the second fusion network, and fuse the second user features of the plurality of third user nodes and the third graph structure features to obtain second fusion features.

[0219] The type identification unit 7043 is configured to invoke the identification network, and identify the first fusion features and the second fusion features to obtain the group type of the target group.

[0220] In another possible implementation manner, the group type identification model further includes a splicing network, referring to Figure 8 The type identification unit 7043 is configured to:

[0221] The splicing network is called to splice the first fusion feature and the second fusion feature, to obtain a splicing feature corresponding to the target group;

[0222] The identification network is called to identify the splicing feature, to obtain a group type of the target group.

[0223] In another possible implementation, the training process of the group type identification model includes:

[0224] obtaining a sample type of a sample group, a sample graph structure feature of a sample graph, and sample user features of a plurality of sample user nodes in a sample target graph, the sample user nodes being nodes corresponding to sample user identifiers, and the sample graph being constructed according to an association relationship between the plurality of sample user identifiers in the sample group;

[0225] The group type identification model is called to identify the sample graph structure feature and the sample user features of the plurality of sample user nodes, to obtain a predicted type of the sample group;

[0226] According to the difference between the sample type and the predicted type, the group type identification model is trained.

[0227] In another possible implementation, referring to Figure 8 , the first screening network 703 is configured to:

[0228] The number of the plurality of first user nodes is multiplied by a reference ratio to obtain a reference number;

[0229] The attention parameters of the plurality of first user nodes are arranged in descending order, and a reference number of attention parameters arranged in front are selected, and the first user nodes corresponding to the selected attention parameters are selected as the plurality of second user nodes.

[0230] In another possible implementation, the graph structure feature includes an association degree feature between any two user nodes in the plurality of user nodes, referring to Figure 8 , the feature acquisition module 701 is configured to:

[0231] obtain a common occurrence frequency of any two user identifiers in the target group, the common occurrence frequency being a frequency of the any two user identifiers publishing content in the target group based on the any two user identifiers in a plurality of reference time periods;

[0232] According to the common occurrence frequency, the association degree feature between the any two user identifiers is determined, and the association degree feature is in a positive correlation relationship with the common occurrence frequency.

[0233] In another possible implementation, the user feature includes a user behavior feature and a user attribute feature, referring to Figure 9 , the feature acquisition module 701 is configured to:

[0234] obtaining a user social network, the user social network comprising a plurality of registered user identities;

[0235] obtaining user behavior features of the plurality of user identities in the target group according to the user social network;

[0236] obtaining user attribute features of the plurality of user identities in the target group according to user portrait information corresponding to the plurality of user identities in the target group.

[0237] All the optional technical solutions described above can be combined to form optional embodiments of the present application, and will not be repeated here.

[0238] It should be noted that: the group type identification device provided in the above embodiments only uses the division of the above functional modules as an example to illustrate the identification of the group type. In actual application, the above functions can be completed by different functional modules according to needs, that is, the internal structure of the computer device is divided into different functional modules to complete all or part of the functions described above. In addition, the group type identification device and the group type identification method provided in the above embodiments belong to the same concept, and the specific implementation process is described in the method embodiments, which will not be repeated here.

[0239] The embodiments of the present application also provide a computer device, which comprises a processor and a memory, and the memory stores at least one computer program, which is loaded and executed by the processor to implement the operations performed in the group type identification method of the above embodiments.

[0240] Optionally, the computer device is provided as a terminal. Figure 9 Fig. 9 is a structural schematic diagram of a terminal 900 provided by the embodiments of the present application. The terminal 900 can be a portable mobile terminal, such as a smart phone, a tablet computer, an MP3 player (Moving Picture Experts Group Audio Layer III), an MP4 player (Moving Picture Experts Group Audio Layer IV), a notebook computer or a desktop computer. The terminal 900 can also be referred to as a user equipment, a portable terminal, a laptop terminal, a desktop terminal, or other names.

[0241] The terminal 900 comprises a processor 901 and a memory 902.

[0242] The processor 901 can include one or more processing cores, such as a 4-core processor, an 8-core processor, and the like. The processor 901 can be implemented in the form of at least one of a DSP (Digital Signal Processing), an FPGA (Field-Programmable Gate Array), a PLA (Programmable Logic Array), and the like. The processor 901 can also include a main processor and a co-processor. The main processor is a processor for processing data in an awake state, also referred to as a CPU (Central Processing Unit). The co-processor is a low-power consumption processor for processing data in a standby state. In some embodiments, the processor 901 can be integrated with a GPU (Graphics Processing Unit) for rendering and drawing content to be displayed by the display screen. In some embodiments, the processor 901 can further include an AI (Artificial Intelligence) processor for processing machine learning-related computing operations.

[0243] The memory 902 can include one or more computer-readable storage media that can be non-transitory. The memory 902 can also include a high-speed random access memory, and a nonvolatile memory such as one or more disk storage devices, flash storage devices. In some embodiments, the non-transitory computer-readable storage medium in the memory 902 is used to store at least one computer program for being executed by the processor 901 to implement the group type identification method provided by the method embodiments in the present application.

[0244] In some embodiments, the terminal 900 can also optionally include a peripheral device interface 903 and at least one peripheral device. The processor 901, the memory 902, and the peripheral device interface 903 can be connected through a bus or a signal line. Each peripheral device can be connected to the peripheral device interface 903 through a bus, a signal line, or a circuit board. Specifically, the peripheral device includes at least one of a radio frequency circuit 904, a display screen 905, a camera component 906, an audio circuit 907, and a power supply 908.

[0245] The peripheral interface 903 can be used to connect at least one I / O (Input / Output) related peripheral device to the processor 901 and the memory 902. In some embodiments, the processor 901, the memory 902 and the peripheral interface 903 are integrated on the same chip or circuit board; in some other embodiments, any one or two of the processor 901, the memory 902 and the peripheral interface 903 can be implemented on a separate chip or circuit board, and the present embodiments are not limited in this regard.

[0246] The radio frequency circuit 904 is configured to receive and send RF (Radio Frequency) signals, also known as electromagnetic signals. The radio frequency circuit 904 communicates with communication networks and other communication devices through electromagnetic signals. The radio frequency circuit 904 converts electrical signals to electromagnetic signals for transmission, or converts electromagnetic signals received to electrical signals. Optionally, the radio frequency circuit 904 includes an antenna system, an RF transceiver, one or more amplifiers, a tuner, an oscillator, a digital signal processor, a codec chipset, a subscriber identity module card, and the like. The radio frequency circuit 904 can communicate with other terminals through at least one wireless communication protocol. The wireless communication protocol includes but is not limited to the World Wide Web, a metropolitan area network, an intranet, various generations of mobile communication networks (2G, 3G, 4G and 5G), a wireless local area network, and / or a WiFi (Wireless Fidelity) network. In some embodiments, the radio frequency circuit 904 can also include NFC (Near Field Communication) related circuitry, and the present application is not limited in this regard.

[0247] The display screen 905 is configured to display a UI (User Interface). The UI can include graphics, text, icons, video, and any combination thereof. When the display screen 905 is a touch display screen, the display screen 905 is further configured to capture touch signals on or above the surface of the display screen 905. The touch signals can be input to the processor 901 as control signals for processing. In this case, the display screen 905 can also be configured to provide virtual buttons and / or virtual keyboard, also known as soft buttons and / or soft keyboard. In some embodiments, the display screen 905 can be one, disposed on the front panel of the terminal 900; in other embodiments, the display screen 905 can be at least two, respectively disposed on different surfaces of the terminal 900 or in a folding design; in other embodiments, the display screen 905 can be a flexible display screen, disposed on a curved surface or a folding surface of the terminal 900. Even, the display screen 905 can also be disposed in an irregular shape, i.e., a special-shaped screen. The display screen 905 can be made of LCD (Liquid Crystal Display), OLED (Organic Light-Emitting Diode), etc.

[0248] The camera assembly 906 is configured to capture images or videos. Optionally, the camera assembly 906 includes a front camera and a rear camera. The front camera is disposed on the front panel of the terminal, and the rear camera is disposed on the back of the terminal. In some embodiments, the rear camera is at least two, which are any one of a main camera, a depth-of-field camera, a wide-angle camera, and a telephoto camera, to realize the background blur function by fusing the main camera and the depth-of-field camera, the panoramic shooting and VR (Virtual Reality) shooting function by fusing the main camera and the wide-angle camera, or other fusion shooting functions. In some embodiments, the camera assembly 906 can further include a flash. The flash can be a single-color temperature flash or a dual-color temperature flash. The dual-color temperature flash refers to the combination of a warm light flash and a cold light flash, which can be used for light compensation under different color temperatures.

[0249] The audio circuit 907 can include a microphone and a speaker. The microphone is used to collect sound waves of the user and the environment, and convert the sound waves into an electrical signal input to the processor 901 for processing, or input to the radio frequency circuit 904 to realize voice communication. For the purpose of stereo sound collection or noise reduction, the microphone can be multiple, respectively arranged at different parts of the terminal 900. The microphone can also be an array microphone or an omnidirectional collection type microphone. The speaker is used to convert the electrical signal from the processor 901 or the radio frequency circuit 904 into sound waves. The speaker can be a conventional diaphragm speaker, or a piezoelectric ceramic speaker. When the speaker is a piezoelectric ceramic speaker, not only can it convert electrical signals into sound waves that humans can hear, but it can also convert electrical signals into sound waves that humans cannot hear for ranging purposes. In some embodiments, the audio circuit 907 can also include a headphone jack.

[0250] The power supply 908 is used to supply power to various components in the terminal 900. The power supply 908 can be alternating current, direct current, disposable battery or rechargeable battery. When the power supply 908 includes a rechargeable battery, the rechargeable battery can be a wired charging battery or a wireless charging battery. The wired charging battery is a battery charged through a wired line, and the wireless charging battery is a battery charged through a wireless coil. The rechargeable battery can also be used to support fast charging technology.

[0251] In some embodiments, the terminal 900 also includes one or more sensors 909. The one or more sensors 909 include, but are not limited to, an acceleration sensor 910, a gyroscope sensor 911, a pressure sensor 912, an optical sensor 913, and a proximity sensor 914.

[0252] The acceleration sensor 910 can detect the acceleration in three coordinate axes of the coordinate system established by the terminal 900. For example, the acceleration sensor 910 can be used to detect the components of gravitational acceleration in three coordinate axes. The processor 901 can control the display screen 905 to display the user interface in a landscape view or a portrait view according to the gravitational acceleration signal collected by the acceleration sensor 910. The acceleration sensor 910 can also be used for game or user motion data collection.

[0253] The gyroscope sensor 911 can detect the body direction and rotation angle of the terminal 900, and the gyroscope sensor 911 can collect 3D actions of the user on the terminal 900 in cooperation with the acceleration sensor 910. The processor 901 can realize the following functions according to the data collected by the gyroscope sensor 911: motion sensing (such as changing the UI according to the user's tilt operation), image stabilization when shooting, game control, and inertial navigation.

[0254] The pressure sensor 912 can be arranged at the side frame of the terminal 900 and / or the lower layer of the display screen 905. When the pressure sensor 912 is arranged at the side frame of the terminal 900, the holding signal of the user to the terminal 900 can be detected, and the left-hand or right-hand recognition or shortcut operation can be performed by the processor 901 according to the holding signal collected by the pressure sensor 912. When the pressure sensor 912 is arranged at the lower layer of the display screen 905, the operable control on the UI interface can be controlled by the processor 901 according to the pressure operation of the user to the display screen 905. The operable control includes at least one of the button control, the scroll bar control, the icon control and the menu control.

[0255] The optical sensor 913 is used to collect the ambient light intensity. In one embodiment, the processor 901 can control the display brightness of the display screen 905 according to the ambient light intensity collected by the optical sensor 913. Specifically, when the ambient light intensity is high, the display brightness of the display screen 905 is increased; when the ambient light intensity is low, the display brightness of the display screen 905 is decreased. In another embodiment, the processor 901 can also dynamically adjust the shooting parameter of the camera assembly 906 according to the ambient light intensity collected by the optical sensor 913.

[0256] The proximity sensor 914, also called the distance sensor, is arranged at the front panel of the terminal 900. The proximity sensor 914 is used to collect the distance between the user and the front of the terminal 900. In one embodiment, when the proximity sensor 914 detects that the distance between the user and the front of the terminal 900 gradually decreases, the display screen 905 is switched from the bright screen state to the screen-off state by the processor 901; when the proximity sensor 914 detects that the distance between the user and the front of the terminal 900 gradually increases, the display screen 905 is switched from the screen-off state to the bright screen state by the processor 901.

[0257] Those skilled in the art can understand that the structure shown in the above embodiments is not a limitation on the terminal 900, and the terminal 900 can include more or less components than the structure shown in the above embodiments, or combine some components, or adopt a different component arrangement. Figure 10

[0258] Optionally, the computer device is provided as a server. ​ ​is a structural schematic diagram of a server provided by an embodiment of the present application. The server 1000 can be quite different due to different configurations and performances, and can include one or more processors (Central Processing Units, CPUs) 1001 and one or more memories 1002. The memory 1002 stores at least one computer program, which is loaded and executed by the processor 1001 to implement the method provided by each method embodiment described above. Of course, the server can also have a wired or wireless network interface, a keyboard, an input and output interface, and other components for implementing device functions, and the like, so as to perform input and output. The server can also include other components for implementing device functions, which are not described here.

[0259] An embodiment of the present application further provides a computer readable storage medium, which stores at least one computer program. The at least one computer program is loaded and executed by a processor to implement the operations performed in the group type identification method of the above embodiment.

[0260] An embodiment of the present application further provides a computer program product or a computer program, which includes computer program code stored in a computer readable storage medium. A processor of a computer device reads the computer program code from the computer readable storage medium, and the processor executes the computer program code, so that the computer device implements the operations performed in the group type identification method of the above embodiment.

[0261] Those skilled in the art can understand that all or part of the steps of the above embodiments can be completed by hardware, or by a program instructing related hardware to complete, and the program can be stored in a computer readable storage medium. The storage medium mentioned above can be a read-only memory, a magnetic disk or an optical disk.

[0262] The above is only an optional embodiment of the present application, and does not limit the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A group type identification method, characterized by, The method comprises: obtaining first graph structure features of a first target graph and first user features of a plurality of first user nodes in the first target graph, the first user nodes being nodes corresponding to user identities in a target group, and the first target graph being constructed according to association relationships between the plurality of first user nodes; processing the first graph structure features and the plurality of first user features to obtain attention parameters corresponding to each first user node in the first target graph, the attention parameters being used to represent importance of the first user nodes in the first target graph; selecting a plurality of second user nodes from the plurality of first user nodes according to the obtained plurality of attention parameters, the plurality of second user nodes corresponding to attention parameters greater than those of other first user nodes; identifying first user features of the plurality of second user nodes and second graph structure features of a second target graph to obtain a group type of the target group, the second target graph being constructed according to association relationships between the plurality of second user nodes; The method further comprises: adjusting the first user features of the plurality of second user nodes according to the second graph structure features to obtain second user features of the plurality of second user nodes; processing the second graph structure features and the plurality of second user features to obtain attention parameters of each second user node in the second target graph; selecting a plurality of third user nodes from the plurality of second user nodes according to the obtained plurality of attention parameters, the plurality of third user nodes corresponding to attention parameters greater than those of other second user nodes; The identifying of the first user features of the plurality of second user nodes and the second graph structure features of the second target graph to obtain the group type of the target group comprises: identifying the first user features of the plurality of second user nodes, the second graph structure features, second user features of the plurality of third user nodes, and third graph structure features of a third target graph to obtain the group type of the target group, the third target graph being constructed according to association relationships between the plurality of third user nodes.

2. The method of claim 1, wherein, The identifying of the first user features of the plurality of second user nodes, the second graph structure features, the second user features of the plurality of third user nodes, and the third graph structure features of the third target graph to obtain the group type of the target group comprises: fusing the first user features of the plurality of second user nodes and the second graph structure features to obtain first fused features; fusing the second user features of the plurality of third user nodes and the third graph structure features to obtain second fused features; identifying the first fused features and the second fused features to obtain the group type of the target group.

3. The method of claim 2, wherein, The fusing of the first user features of the plurality of second user nodes and the second graph structure features to obtain first fused features comprises: averaging the first user features of the plurality of second user nodes and the second graph structure feature according to the number of the second user nodes, to obtain an average user feature corresponding to the plurality of second user nodes; concatenating the average user feature and a maximum user feature in the first user features of the plurality of second user nodes, to obtain the first fusion feature.

4. The method of claim 2, wherein, The identifying the first fusion feature and the second fusion feature to obtain the group type of the target group comprises: concatenating the first fusion feature and the second fusion feature to obtain a concatenated feature corresponding to the target group; identifying the concatenated feature to obtain the group type of the target group.

5. The method of claim 1, wherein, The group type identification model comprises a first attention network, a first screening network and an identification network, and the processing the first graph structure feature and the plurality of first user features to obtain the attention parameter corresponding to each first user node in the first target graph comprises: calling the first attention network to process the first graph structure feature and the plurality of first user features to obtain the attention parameter corresponding to each first user node in the first target graph; the selecting the plurality of second user nodes from the plurality of first user nodes according to the obtained plurality of attention parameters comprises: calling the first screening network to select the plurality of second user nodes from the plurality of first user nodes according to the obtained plurality of attention parameters; the identifying the first user features of the plurality of second user nodes and the second graph structure feature of the second target graph to obtain the group type of the target group comprises: calling the identification network to identify the first user features of the plurality of second user nodes and the second graph structure feature to obtain the group type of the target group.

6. The method of claim 5, wherein, The group type identification model further comprises a first convolution network, a second attention network and a second screening network, and after the calling the first screening network to select the plurality of second user nodes from the plurality of first user nodes according to the obtained plurality of attention parameters, the method further comprises: calling the first convolution network to adjust the first user features of the plurality of second user nodes according to the second graph structure feature to obtain second user features of the plurality of second user nodes; calling the second attention network to process the second graph structure feature and the plurality of second user features to obtain the attention parameter of each second user node in the second target graph; calling the second screening network to select a plurality of third user nodes from the plurality of second user nodes according to the obtained plurality of attention parameters, the attention parameter corresponding to the plurality of third user nodes being greater than the attention parameter corresponding to other second user nodes.

7. The method of claim 6, wherein, The calling the identification network to identify the first user features of the plurality of second user nodes and the second graph structure feature to obtain the group type of the target group comprises: calling the identification network to identify the first user features of the plurality of second user nodes, the second graph structure features, the second user features of the plurality of third user nodes, and the third graph structure features of the third target graph, to obtain the group type of the target group, the third target graph being constructed according to the association relationships between the plurality of third user nodes.

8. The method of claim 7, wherein, The group type identification model further includes a first fusion network and a second fusion network, and the calling the identification network to identify the first user features of the plurality of second user nodes, the second graph structure features, the second user features of the plurality of third user nodes, and the third graph structure features of the third target graph, to obtain the group type of the target group, includes: calling the first fusion network to fuse the first user features of the plurality of second user nodes and the second graph structure features, to obtain first fusion features; calling the second fusion network to fuse the second user features of the plurality of third user nodes and the third graph structure features, to obtain second fusion features; calling the identification network to identify the first fusion features and the second fusion features, to obtain the group type of the target group.

9. The method of claim 8, wherein, The group type identification model further includes a splicing network, and the calling the identification network to identify the first fusion features and the second fusion features, to obtain the group type of the target group, includes: calling the splicing network to splice the first fusion features and the second fusion features, to obtain splicing features corresponding to the target group; calling the identification network to identify the splicing features, to obtain the group type of the target group.

10. The method according to any one of claims 5 to 9, characterized in that, The training process of the group type identification model includes: obtaining a sample type of a sample group, sample graph structure features of a sample graph, and sample user features of a plurality of sample user nodes in the sample graph, the sample user nodes being nodes corresponding to sample user identifiers, and the sample graph being constructed according to association relationships between the plurality of sample user identifiers in the sample group; calling the group type identification model to identify the sample graph structure features and the sample user features of the plurality of sample user nodes, to obtain a predicted type of the sample group; training the group type identification model according to differences between the sample type and the predicted type.

11. The method of claim 1, wherein, The selecting a plurality of second user nodes from the plurality of first user nodes according to the obtained plurality of attention parameters includes: multiplying the number of the plurality of first user nodes by a reference ratio to obtain a reference number; arranging the attention parameters of the plurality of first user nodes in descending order, selecting a reference number of attention parameters arranged in front, and taking the first user nodes corresponding to the selected attention parameters as the selected plurality of second user nodes.

12. The method of claim 1, wherein, The graph structure features include association degree features between any two user nodes in the plurality of user nodes, and the obtaining the first graph structure features of the first target graph and the first user features of the plurality of first user nodes in the first target graph includes: obtaining a common occurrence number of any two user identifiers in the target group, the common occurrence number being a number of times that the any two user identifiers jointly publish content in the target group based on a plurality of reference time periods; determining a correlation degree feature between the any two user identifiers according to the common occurrence number, the correlation degree feature being in a positive correlation relationship with the common occurrence number.

13. The method of claim 1, wherein, The user features include user behavior features and user attribute features, and the obtaining of the first graph structure feature of the first target graph and the first user features of a plurality of first user nodes in the first target graph includes: obtaining a user social network, the user social network including a plurality of registered user identifiers; obtaining user behavior features of a plurality of user identifiers in the target group according to the user social network; obtaining user attribute features of a plurality of user identifiers in the target group according to user portrait information corresponding to the plurality of user identifiers in the target group.

14. A group type identifying apparatus characterized by comprising: The device includes: a feature obtaining module configured to obtain a first graph structure feature of a first target graph and first user features of a plurality of first user nodes in the first target graph, the first user nodes being nodes corresponding to user identifiers in a target group, and the first target graph being constructed according to correlation relationships between the plurality of first user nodes; a first attention obtaining module configured to process the first graph structure feature and the plurality of first user features to obtain an attention parameter corresponding to each first user node in the first target graph, the attention parameter being used to represent an importance of the first user node in the first target graph; a first screening module configured to select a plurality of second user nodes from the plurality of first user nodes according to the obtained plurality of attention parameters, the plurality of second user nodes corresponding to attention parameters greater than those of other first user nodes; a type identifying module configured to identify the first user features of the plurality of second user nodes and a second graph structure feature of a second target graph to obtain a group type of the target group, the second target graph being constructed according to correlation relationships between the plurality of second user nodes; The device further includes: a feature adjusting module configured to adjust the first user features of the plurality of second user nodes according to the second graph structure feature to obtain second user features of the plurality of second user nodes; a second attention obtaining module configured to process the second graph structure feature and the plurality of second user features to obtain an attention parameter of each second user node in the second target graph; a second screening module configured to select a plurality of third user nodes from the plurality of second user nodes according to the obtained plurality of attention parameters, the plurality of third user nodes corresponding to attention parameters greater than those of other second user nodes. The type identifying module is configured to identify the first user features of the plurality of second user nodes, the second graph structure features, the second user features of the plurality of third user nodes, and the third graph structure features of the third target graph to obtain the group type of the target group, the third target graph being constructed according to the association relationships among the plurality of third user nodes.

15. The apparatus of claim 14, wherein, The type identifying module comprises: A first fusion unit configured to fuse the first user features of the plurality of second user nodes and the second graph structure features to obtain first fusion features; A second fusion unit configured to fuse the second user features of the plurality of third user nodes and the third graph structure features to obtain second fusion features; A type identifying unit configured to identify the first fusion features and the second fusion features to obtain the group type of the target group.

16. The apparatus of claim 15, wherein, The first fusion unit is configured to: perform average processing on the first user features of the plurality of second user nodes and the second graph structure features according to the number of the second user nodes to obtain average user features corresponding to the plurality of second user nodes; and splice the average user features and maximum user features in the first user features of the plurality of second user nodes to obtain the first fusion features.

17. The apparatus of claim 15, wherein, The type identifying unit is configured to: splice the first fusion features and the second fusion features to obtain spliced features corresponding to the target group; and identify the spliced features to obtain the group type of the target group.

18. The apparatus of claim 14, wherein, The group type identifying model comprises a first attention network, a first screening network, and an identifying network, The first attention obtaining module is configured to invoke the first attention network to process the first graph structure features and the plurality of first user features to obtain attention parameters corresponding to each first user node in the first target graph; The first screening module is configured to invoke the first screening network to select the plurality of second user nodes from the plurality of first user nodes according to the plurality of obtained attention parameters; The type identifying module is configured to invoke the identifying network to identify the first user features of the plurality of second user nodes and the second graph structure features to obtain the group type of the target group.

19. The apparatus of claim 18, wherein, The group type identifying model further comprises a first convolution network, a second attention network, and a second screening network; The feature adjusting module is configured to invoke the first convolution network to adjust the first user features of the plurality of second user nodes according to the second graph structure features to obtain second user features of the plurality of second user nodes; The second attention obtaining module is configured to invoke the second attention network to process the second graph structure features and the plurality of second user features to obtain attention parameters of each second user node in the second target graph; The second screening module is configured to call the second screening network, and select a plurality of third user nodes from the plurality of second user nodes according to the obtained plurality of attention parameters, the plurality of third user nodes corresponding to attention parameters greater than those of other second user nodes.

20. The apparatus of claim 19, wherein, The type identification module is configured to call the identification network, and identify the first user features of the plurality of second user nodes, the second graph structure features, the second user features of the plurality of third user nodes, and the third graph structure features of the third target graph to obtain the group type of the target group, the third target graph being constructed according to the association relationships between the plurality of third user nodes.

21. The apparatus of claim 20, wherein, The group type identification model further includes a first fusion network and a second fusion network, and the type identification module includes: A first fusion unit configured to call the first fusion network, and fuse the first user features of the plurality of second user nodes and the second graph structure features to obtain first fusion features; A second fusion unit configured to call the second fusion network, and fuse the second user features of the plurality of third user nodes and the third graph structure features to obtain second fusion features; A type identification unit configured to call the identification network, and identify the first fusion features and the second fusion features to obtain the group type of the target group.

22. The apparatus of claim 21, wherein, The group type identification model further includes a splicing network, and the type identification unit is configured to: Call the splicing network to splice the first fusion features and the second fusion features to obtain splicing features corresponding to the target group; and Call the identification network to identify the splicing features to obtain the group type of the target group.

23. The apparatus of any of claims 18-22, wherein, The training process of the group type identification model includes: Obtaining a sample type of a sample group, sample graph structure features of a sample graph, and sample user features of a plurality of sample user nodes in the sample graph, the sample user nodes being nodes corresponding to sample user identifiers, and the sample graph being constructed according to association relationships between the plurality of sample user identifiers in the sample group; Calling the group type identification model to identify the sample graph structure features and the sample user features of the plurality of sample user nodes to obtain a predicted type of the sample group; Training the group type identification model according to differences between the sample type and the predicted type.

24. The apparatus of claim 14, wherein, The first screening module is configured to: Multiply the number of the plurality of first user nodes by a reference ratio to obtain a reference number; Arrange the attention parameters of the plurality of first user nodes in descending order, select a reference number of attention parameters arranged in the front, and select the first user nodes corresponding to the selected attention parameters as the plurality of second user nodes.

25. The apparatus of claim 14, wherein, The graph structure features include association degree features between any two user nodes in the plurality of user nodes, and the feature acquisition module is configured to: obtaining a common occurrence number of any two user identifiers in the target group, the common occurrence number being a number of times that the any two user identifiers jointly publish content in the target group based on a plurality of reference time periods; determining a correlation degree feature between the any two user identifiers according to the common occurrence number, the correlation degree feature being in a positive correlation relationship with the common occurrence number.

26. The apparatus of claim 14, wherein, The user features include user behavior features and user attribute features, and the feature obtaining module is configured to: obtain a user social network, the user social network including a plurality of registered user identifiers; obtain user behavior features of the plurality of user identifiers in the target group according to the user social network; obtain user attribute features of the plurality of user identifiers in the target group according to user portrait information corresponding to the plurality of user identifiers in the target group.

27. A computer device, comprising: The computer device includes a processor and a memory, and the memory stores at least one computer program, which is loaded and executed by the processor to implement the operations performed in the group type identification method according to any one of claims 1 to 13.

28. A computer-readable storage medium, characterized in that, The computer readable storage medium stores at least one computer program, which is loaded and executed by the processor to implement the operations performed in the group type identification method according to any one of claims 1 to 13.

29. A computer program product, characterised in that, The computer program product includes computer program code stored in a computer readable storage medium, and the processor of the computer device reads the computer program code from the computer readable storage medium, and the processor executes the computer program code, so that the computer device implements the operations performed in the group type identification method according to any one of claims 1 to 13.

Images