Social network group detection method, apparatus and device, medium and program product

Abstract

The embodiment of the invention provides a social network group detection method and device, equipment, a medium and a program product, and the method can comprise the steps: obtaining the transaction data volume of each transaction between objects in a target group, and then carrying out the transaction data volume of each transaction between the objects for each transaction initiator in each object, calculating a first total transaction data volume and a second total transaction data volume, then taking a transaction initiator and a transaction receiver with transactions in each object as an object group, and for each object group, calculating a ratio between the first total transaction data volume and the second total transaction data volume corresponding to the object group; and taking the ratio corresponding to each object group as the influence degree of the transaction initiator in the object group on the transaction receiver, calculating the transaction importance degree of each object in the target group for the ratio corresponding to each object group in the target group, and performing group classification processing on the target group based on the transaction importance degree, and obtaining the probability that the target group belongs to the group of the target type.

Description

Technical Field

[0001] This application belongs to the field of computer technology, and more specifically, it relates to a method, apparatus, device, medium, and program product for detecting social network groups. Background Technology

[0002] With the development of computer technology and social applications, various types of groups can be established based on these applications, such as family groups, work groups, and food groups. Furthermore, with the expansion of payment services, group transactions can be conducted within these groups. Group transactions, in particular, are frequently exploited by criminal organizations, and due to the nature of groups, assessing transaction risks is quite difficult. Therefore, it is necessary to monitor groups established based on social applications to determine whether they belong to specific types of groups for focused monitoring, thus preventing harm to the interests of users within these groups.

[0003] Therefore, how to detect social network groups to determine whether these social network groups belong to the target type has become a key issue. Summary of the Invention

[0004] The purpose of this application is to provide a method, apparatus, device, medium, and program product for detecting social network groups to determine whether these social network groups belong to a target type of group. To achieve the above objective, the technical solution provided by this application is as follows: Firstly, a method for detecting social network groups is provided, including: Obtain the transaction data volume of each transaction between each object in the target group; For each transaction initiator among the various objects, based on the transaction data volume of each transaction initiated by the transaction initiator, a first total transaction data volume between the transaction initiator and each transaction recipient corresponding to the transaction initiator is calculated, and a second total transaction data volume of the transactions initiated by the transaction initiator is calculated. Each of the objects is divided into an object group by a transaction initiator and a transaction receiver. For each object group, the ratio between the first total transaction data volume corresponding to the object group and the second total transaction volume corresponding to the transaction initiator in the object group is calculated. The ratio corresponding to each object group is used as the degree of influence of the transaction initiator on the transaction recipient in that object group. For the ratio corresponding to each object group in the target group, the transaction importance of each object in the target group in the target group is calculated. Based on the transaction importance of each object in the target group, the target group is classified using a trained classification model to obtain the probability that the target group belongs to the target type group.

[0005] In one possible implementation, calculating the transaction importance of each object in the target group within the target group, based on the ratio corresponding to each object group in the target group, includes: Based on the first number of objects in the target group and the ratio corresponding to each object group, a first matrix is ​​constructed in which the number of rows and columns are equal to the first number, and the element value at any position is equal to the ratio corresponding to the object group at that position; wherein, the number of rows and columns of the first matrix are both the number of objects contained in the target group, and the element at any position in the first matrix is ​​the ratio corresponding to the object group represented by that position. An initial feature vector is constructed, and based on the initial feature vector and the first matrix, the feature vector of the first matrix is ​​calculated, wherein the feature vector of the first matrix is ​​used to characterize the transaction importance of each object in the target group in the target group.

[0006] In another possible implementation, the step of classifying the target group based on the transaction importance of each object in the target group using a trained classification model to obtain the probability that the target group belongs to a target type group includes: The eigenvectors of the first matrix are sorted according to a preset order to obtain sorted eigenvectors. Based on the sorted feature vectors, and through the trained classification model, the probability that the target group belongs to the target type group is obtained.

[0007] In another possible implementation, calculating the transaction importance of each object in the target group within the target group, based on the ratio corresponding to each object group in the target group, includes: Obtain the target transaction reason directed graph corresponding to the target transaction group; wherein, a node in the target transaction reason directed graph represents an object, there is an edge between the nodes corresponding to the two objects that have a transaction, the direction of the edge is from the node of the transaction initiator to the node of the transaction receiver, the weight of the edge is used to represent the reliability of the transaction relationship, and the weight of the edge is directly proportional to the reliability of the transaction relationship. Take one transaction initiator and one transaction receiver in each of the objects as an object group. For each object group in the target group, determine the weight of the connecting edge corresponding to the object group from the directed graph of the target transaction cause. Based on the weight of the edges corresponding to the object group, the ratio between the first total transaction data volume corresponding to the object group and the second total transaction volume corresponding to the transaction initiator in the object group is adjusted. The weight of each edge in the directed graph of the transaction reason is inversely proportional to the ratio of the object group corresponding to the edge. Based on the adjusted ratios of each object group, the transaction importance of each object in the target group is calculated within the target group.

[0008] In another possible implementation, the method for obtaining the directed graph of the target transaction reasons corresponding to the target transaction group includes: Obtain the relationship graph between each object in the target group; wherein, the weight of the edge between any two objects represents the degree of intimacy between the two objects, and the weight of the edge between any two objects is directly proportional to the degree of intimacy between the two objects; Obtain chat data between various objects in the target group, and extract context chat data corresponding to the transaction between each object from the chat data; Based on the contextual chat data corresponding to the transactions of the various objects, a first directed graph of transaction reasons is constructed; wherein, a node in the first directed graph of transaction reasons represents an object, there is an edge between the nodes corresponding to the two objects that have a transaction, the direction of the edge is from the node of the transaction initiator to the node of the transaction receiver, the weight of the edge is used to represent the reliability of the transaction relationship, and the weight of the edge is directly proportional to the reliability of the transaction relationship. Based on the relationship graph between the objects in the target group, the weights of each edge in the first directed graph of transaction reasons are adjusted to obtain the second directed graph of transaction reasons. At least one of the first directed graph of transaction reasons and the second directed graph of transaction reasons is determined as the directed graph of target transaction reasons corresponding to the target transaction group.

[0009] In another possible implementation, the second total transaction data volume corresponding to each transaction initiator in the respective objects is obtained in the following way: Each object in the target group is treated as a node. An edge is constructed between the nodes corresponding to the objects that have a transaction, pointing from the node of the transaction initiator to the node of the transaction receiver. The first total transaction data volume between the objects is used as the weight to construct a directed graph of the transaction relationship corresponding to the target group. For each source node in the directed graph of the transaction relationship, the second total transaction data volume of the transaction initiator corresponding to the source node is obtained by calculating the sum of the weights of all the edges sent by the source node.

[0010] In another possible implementation, before calculating the ratio between the first total transaction data volume corresponding to the object group and the second total transaction volume corresponding to the transaction initiator in the object group, the method further includes: Determine that the second total transaction volume corresponding to the transaction initiator in the object group is greater than or equal to a preset transaction volume threshold. The method further includes: If the total transaction volume corresponding to the transaction initiator in the object group is less than the preset transaction volume threshold, then the ratio between the first total transaction volume corresponding to the object group and the preset transaction volume threshold is calculated, and this ratio is used as the ratio between the first total transaction data volume corresponding to the object group and the second total transaction volume corresponding to the transaction initiator in the object group.

[0011] In another possible implementation, the transaction data volume of a single transaction includes at least one of the following: transaction resource volume and number of transactions; If the transaction data volume of a single transaction includes: transaction resource volume and transaction frequency, then, for each object group, calculating the ratio between the first total transaction data volume corresponding to that object group and the second total transaction volume corresponding to the transaction initiator in that object group includes: Calculate the first ratio between the first total transaction resource amount corresponding to the object group and the second total transaction resource amount corresponding to the transaction initiator in the object group; Calculate the second ratio between the first total number of transactions corresponding to the object group and the second total number of transactions corresponding to the transaction initiators in the object group; Obtain the target weight information corresponding to the first ratio and the second ratio respectively; Based on the first ratio, the second ratio, and their respective weight information, the ratio between the first total transaction data volume corresponding to the object group and the second total transaction volume corresponding to the transaction initiator in the object group is calculated.

[0012] In another possible implementation, for each group of objects, obtaining the target weight information corresponding to the first ratio and the second ratio respectively includes: Obtain the first weight information corresponding to the first ratio and the second ratio in the object group. The first weight information is obtained by weighting the number of objects contained in the target group, the first ratio, and the second ratio through a weight model. Obtain the transaction habit characteristics of the transaction initiators in the object group, as well as the time interval between two adjacent transactions in the object group; Based on the transaction habit characteristics and the time interval between two adjacent transactions, the first weight information is adjusted to obtain the target weight information corresponding to the first ratio and the second ratio respectively.

[0013] In another possible implementation, if the probability that the target group belongs to the target type group is greater than a preset probability threshold, the method further includes: Objects in the target group whose transaction importance is greater than a preset importance threshold are identified as target objects.

[0014] In another possible implementation, the trained classification model is obtained by training the initial model in the following way: Multiple training samples are obtained, each training sample including: the amount of transaction data between each object in a sample group, and the label corresponding to the sample group. The label corresponding to a sample group is used to characterize whether the sample group belongs to the target type group. For each sample group, obtain the transaction data volume of each transaction between each object in that sample group; For each transaction initiator among the various objects, based on the transaction data volume of each transaction initiated by the transaction initiator, a first total transaction data volume between the transaction initiator and each transaction recipient corresponding to the transaction initiator is calculated, and a second total transaction data volume of the transactions initiated by the transaction initiator is calculated. Each of the objects is divided into an object group by a transaction initiator and a transaction receiver. For each object group, the ratio between the first total transaction data volume corresponding to the sample object group and the second total transaction volume corresponding to the transaction initiator in the object group is calculated. The ratio corresponding to each of the sample object groups is used as the degree of influence of the transaction initiator on the transaction recipient in that sample object group. For the ratio corresponding to each object group in the sample group, the transaction importance of each object in the sample group is calculated in the sample group. Based on the importance of each sample object in the transaction within each sample group and the label corresponding to that sample group, the initial model is trained to obtain the trained classification model.

[0015] Secondly, a device for detecting social network groups is provided, comprising: The acquisition module is used to acquire the transaction data volume of each transaction between various objects in the target group; The first calculation module is used to calculate, for each transaction initiator among the various objects, a first total transaction data volume between the transaction initiator and each transaction receiver corresponding to the transaction initiator, based on the transaction data volume of each transaction initiated by the transaction initiator, and to calculate a second total transaction data volume of the transactions initiated by the transaction initiator. The second calculation module is used to group one transaction initiator and one transaction receiver in each of the objects into an object group. For each object group, the module calculates the ratio between the first total transaction data volume corresponding to the object group and the second total transaction volume corresponding to the transaction initiator in the object group. The third calculation module is used to take the ratio corresponding to each object group as the degree of influence of the transaction initiator on the transaction recipient in that object group, and to calculate the transaction importance of each object in the target group in the target group for the ratio corresponding to each object group in the target group. The group classification processing module is used to classify the target group based on the transaction importance of each object in the target group using a trained classification model, and to obtain the probability that the target group belongs to the target type of group.

[0016] In one possible implementation, when the third calculation module calculates the transaction importance of each object in the target group based on the ratio corresponding to each object group in the target group, it is specifically used for: Based on the first number of objects in the target group and the ratio corresponding to each object group, a first matrix is ​​constructed in which the number of rows and columns are equal to the first number, and the element value at any position is equal to the ratio corresponding to the object group at that position; wherein, the number of rows and columns of the first matrix are both the number of objects contained in the target group, and the element at any position in the first matrix is ​​the ratio corresponding to the object group represented by that position. An initial feature vector is constructed, and based on the initial feature vector and the first matrix, the feature vector of the first matrix is ​​calculated, wherein the feature vector of the first matrix is ​​used to characterize the transaction importance of each object in the target group in the target group.

[0017] In another possible implementation, when the group classification processing module performs group classification processing on the target group based on the transaction importance of each object in the target group using a trained classification model to obtain the probability that the target group belongs to the target type group, it is specifically used for: The eigenvectors of the first matrix are sorted according to a preset order to obtain sorted eigenvectors. Based on the sorted feature vectors, and through the trained classification model, the probability that the target group belongs to the target type group is obtained.

[0018] In another possible implementation, when the third calculation module calculates the transaction importance of each object in the target group based on the ratio corresponding to each object group in the target group, it is specifically used for: Obtain the target transaction reason directed graph corresponding to the target transaction group; wherein, a node in the target transaction reason directed graph represents an object, there is an edge between the nodes corresponding to the two objects that have a transaction, the direction of the edge is from the node of the transaction initiator to the node of the transaction receiver, the weight of the edge is used to represent the reliability of the transaction relationship, and the weight of the edge is directly proportional to the reliability of the transaction relationship. Take one transaction initiator and one transaction receiver in each of the objects as an object group. For each object group in the target group, determine the weight of the connecting edge corresponding to the object group from the directed graph of the target transaction cause. Based on the weight of the edges corresponding to the object group, the ratio between the first total transaction data volume corresponding to the object group and the second total transaction volume corresponding to the transaction initiator in the object group is adjusted. The weight of each edge in the directed graph of the transaction reason is inversely proportional to the ratio of the object group corresponding to the edge. Based on the adjusted ratios of each object group, the transaction importance of each object in the target group is calculated within the target group.

[0019] In another possible implementation, the method for obtaining the directed graph of the target transaction reasons corresponding to the target transaction group includes: Obtain the relationship graph between each object in the target group; wherein, the weight of the edge between any two objects represents the degree of intimacy between the two objects, and the weight of the edge between any two objects is directly proportional to the degree of intimacy between the two objects; Obtain chat data between various objects in the target group, and extract context chat data corresponding to the transaction between each object from the chat data; Based on the contextual chat data corresponding to the transactions of the various objects, a first directed graph of transaction reasons is constructed; wherein, a node in the first directed graph of transaction reasons represents an object, there is an edge between the nodes corresponding to the two objects that have a transaction, the direction of the edge is from the node of the transaction initiator to the node of the transaction receiver, the weight of the edge is used to represent the reliability of the transaction relationship, and the weight of the edge is directly proportional to the reliability of the transaction relationship. Based on the relationship graph between the objects in the target group, the weights of each edge in the first directed graph of transaction reasons are adjusted to obtain the second directed graph of transaction reasons. At least one of the first directed graph of transaction reasons and the second directed graph of transaction reasons is determined as the directed graph of target transaction reasons corresponding to the target transaction group.

[0020] In another possible implementation, the second total transaction data volume corresponding to each transaction initiator in the respective objects is obtained in the following way: Each object in the target group is treated as a node. An edge is constructed between the nodes corresponding to the objects that have a transaction, pointing from the node of the transaction initiator to the node of the transaction receiver. The first total transaction data volume between the objects is used as the weight to construct a directed graph of the transaction relationship corresponding to the target group. For each source node in the directed graph of the transaction relationship, the second total transaction data volume of the transaction initiator corresponding to the source node is obtained by calculating the sum of the weights of all the edges sent by the source node.

[0021] In another possible implementation, the apparatus includes: a first determining module, wherein, The first determining module is used to determine that the second total transaction volume corresponding to the transaction initiator in the object group is greater than or equal to a preset transaction volume threshold. The device further includes: a fourth computing module, wherein, The fourth calculation module is used to calculate the ratio between the first total transaction volume of the object group and the preset transaction volume threshold when the total transaction volume corresponding to the transaction initiator in the object group is less than the preset transaction volume threshold, and use the ratio as the ratio between the first total transaction data volume of the object group and the second total transaction volume corresponding to the transaction initiator in the object group.

[0022] In another possible implementation, the transaction data volume of a single transaction includes at least one of the following: transaction resource volume and number of transactions; If the transaction data volume of a single transaction includes: transaction resource volume and transaction count, then when the second calculation module calculates the ratio between the first total transaction data volume corresponding to each object group and the second total transaction volume corresponding to the transaction initiator in that object group, it is specifically used for: Calculate the first ratio between the first total transaction resource amount corresponding to the object group and the second total transaction resource amount corresponding to the transaction initiator in the object group; Calculate the second ratio between the first total number of transactions corresponding to the object group and the second total number of transactions corresponding to the transaction initiators in the object group; Obtain the target weight information corresponding to the first ratio and the second ratio respectively; Based on the first ratio, the second ratio, and their respective weight information, the ratio between the first total transaction data volume corresponding to the object group and the second total transaction volume corresponding to the transaction initiator in the object group is calculated.

[0023] In another possible implementation, for each object group, the second calculation module, when obtaining the target weight information corresponding to the first ratio and the second ratio respectively, is specifically used for: Obtain the first weight information corresponding to the first ratio and the second ratio in the object group. The first weight information is obtained by weighting the number of objects contained in the target group, the first ratio, and the second ratio through a weight model. Obtain the transaction habit characteristics of the transaction initiators in the object group, as well as the time interval between two adjacent transactions in the object group; Based on the transaction habit characteristics and the time interval between two adjacent transactions, the first weight information is adjusted to obtain the target weight information corresponding to the first ratio and the second ratio respectively.

[0024] In another possible implementation, if the probability that the target group belongs to the target type group is greater than a preset probability threshold, the device further includes: a second determining module, wherein, The second determining module is used to determine the objects in the target group whose transaction importance is greater than a preset importance threshold as target objects.

[0025] In another possible implementation, the trained classification model is obtained by training the initial model in the following way: Multiple training samples are obtained, each training sample including: the amount of transaction data between each object in a sample group, and the label corresponding to the sample group. The label corresponding to a sample group is used to characterize whether the sample group belongs to the target type group. For each sample group, obtain the transaction data volume of each transaction between each object in that sample group; For each transaction initiator among the various objects, based on the transaction data volume of each transaction initiated by the transaction initiator, a first total transaction data volume between the transaction initiator and each transaction recipient corresponding to the transaction initiator is calculated, and a second total transaction data volume of the transactions initiated by the transaction initiator is calculated. Each of the objects is divided into an object group by a transaction initiator and a transaction receiver. For each object group, the ratio between the first total transaction data volume corresponding to the sample object group and the second total transaction volume corresponding to the transaction initiator in the object group is calculated. The ratio corresponding to each of the sample object groups is used as the degree of influence of the transaction initiator on the transaction recipient in that sample object group. For the ratio corresponding to each object group in the sample group, the transaction importance of each object in the sample group is calculated in the sample group. Based on the importance of each sample object in the transaction within each sample group and the label corresponding to that sample group, the initial model is trained to obtain the trained classification model.

[0026] Thirdly, embodiments of this application also provide an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the social network group detection method provided in any possible implementation of the first aspect.

[0027] Fourthly, embodiments of this application also provide a computer-readable storage medium storing a computer program that, when executed by a processor, implements the social network group detection method provided by any possible implementation of the first aspect.

[0028] Fifthly, embodiments of this application also provide a computer program product, which includes a computer program that, when executed by a processor, implements the method for detecting social network groups provided in any possible implementation of the first aspect.

[0029] The beneficial effects of the technical solution provided in this application are as follows: This application provides a method, apparatus, electronic device, storage medium, and program product for detecting social network groups. In this application embodiment, by acquiring the transaction data volume of each transaction between objects in a target group, and for each transaction initiator in each object, a first total transaction data volume between the transaction initiator and each corresponding transaction recipient, and a second total transaction data volume of transactions initiated by the transaction initiator, can be calculated. Since a transaction initiator and a transaction recipient in each object are considered as an object group, the ratio between the first total transaction data volume and the second total transaction data volume corresponding to each object group can be obtained based on the first and second total transaction data volumes. This ratio serves as the degree of influence of the transaction initiator on the transaction recipient in the object group. Based on this ratio, the transaction importance of each object in the target group can be calculated. Based on the transaction importance of each object in the target group, a trained classification model can be used to obtain the probability that the target group belongs to a target type of group, thereby determining whether the target group belongs to the target type of group. Attached Figure Description

[0030] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments of this application will be briefly introduced below.

[0031] Figure 1 This is a schematic diagram illustrating an application scenario of a social network group detection method according to an embodiment of this application; Figure 2 This is a schematic diagram of a method for detecting social network groups in an embodiment of this application; Figure 3 This is a schematic diagram of a directed graph of a transaction network in an embodiment of this application; Figure 4 This is a schematic diagram of the structure of a fully connected neural network in an embodiment of this application; Figure 5 This is a flowchart illustrating a training method in an embodiment of this application; Figure 6 This is a flowchart illustrating a method for detecting social network groups according to an embodiment of this application. Figure 7 This is a flowchart illustrating another method for detecting social network groups in this application. Figure 8 This is a schematic diagram of a device structure for detecting social network groups according to an embodiment of this application; Figure 9 This is a schematic diagram of the structure of an electronic device according to an embodiment of this application. Detailed Implementation

[0032] The embodiments of this application are described below with reference to the accompanying drawings. It should be understood that the embodiments described below with reference to the accompanying drawings are exemplary descriptions for explaining the technical solutions of the embodiments of this application, and do not constitute a limitation on the technical solutions of the embodiments of this application.

[0033] Those skilled in the art will understand that, unless specifically stated otherwise, the singular forms “a,” “an,” “the,” and “the” used herein may also include the plural forms. It should be further understood that the terms “comprising” and “including” as used in embodiments of this application mean that the corresponding feature can be implemented as the presented feature, information, data, step, operation, element, and / or component, but do not exclude implementation as other features, information, data, step, operation, element, component, and / or combinations thereof supported by the art. It should be understood that when we say that an element is “connected” or “coupled” to another element, the one element can be directly connected or coupled to the other element, or it can mean that the one element and the other element establish a connection relationship through an intermediate element. Furthermore, “connected” or “coupled” as used herein can include wireless connection or wireless coupling. The term “and / or” as used herein indicates at least one of the items defined by the term; for example, “A and / or B” can be implemented as “A,” or as “B,” or as “A and B.” When describing multiple (two or more) items, if the relationship between the multiple items is not explicitly defined, the multiple items can refer to one, several or all of the multiple items. For example, the description of "parameter A includes A1, A2, A3" can be implemented as parameter A includes A1 or A2 or A3, or it can be implemented as parameter A includes at least two of the three items A1, A2 and A3.

[0034] Before providing a further detailed description of the embodiments of this application, the nouns and terms involved in the embodiments of this application will be explained, and the nouns and terms involved in the embodiments of this application shall be interpreted as follows.

[0035] Social Groups: Groups of multiple users who can communicate within them. These social groups can be found in any chat software, online forums, microblogging websites, gaming communities, instant messaging applications, etc.

[0036] Social payments: Personal payments made to individual users via instant messaging (including but not limited to friend transfers, QR code transfers, sending red envelopes, etc.); Social payment fraud: inducing users to make payments in social payment scenarios on social platforms such as instant messaging; Malicious groups: Groups created within applications (such as instant messaging applications) by black market operators or malicious users for the purpose of carrying out fraudulent activities; PageRank: Originally a technique used to measure the importance of a website or webpage, it is now often seen as an unsupervised learning method based on graph link analysis.

[0037] The purpose of this application is to detect social network groups to determine whether these groups belong to a target type, such as malicious groups. Malicious group mining is essentially an application of malicious group discovery, which refers to identifying groups or organizations with malicious or fraudulent behavior from data. This type of mining has important applications in fields such as finance, social media, cybersecurity, and recommendation systems.

[0038] Several algorithms for detecting malicious groups are provided in related technologies: 1. Graph-based methods: Malicious group discovery can be represented as a graph analysis problem. By modeling users within a group and the relationships between them as a graph structure, various graph algorithms (such as community detection) can be applied to identify potential malicious groups. Examples include Louvain community detection. 2. Association rule-based methods: Association rule-based mining methods (such as the Apriori algorithm and FP-growth algorithm) can discover potential malicious groups based on group transaction records; 3. Clustering-based methods: Clustering algorithms (such as K-means, DBSCAN, hierarchical clustering, etc.) can group entities with similar behavioral characteristics into the same category. Focus on clusters with malicious behavioral characteristics to identify potential malicious groups. 4. Representation Learning Methods: Low-dimensional representations of entities and relationships are extracted using deep learning and neural networks (such as autoencoders, graph convolutional networks, etc.). Applying classification or clustering algorithms to the extracted features can effectively detect potential malicious groups; The methods used to determine whether these social network groups belong to the target type (e.g., malicious groups) through the aforementioned technologies may have the following problems: 1. Graph-based methods: The computational cost of graph algorithms can increase significantly with the number of nodes and edges. For large-scale graphs, this can lead to computational and memory limitations. 2. Association rule-based methods: Association rule mining primarily focuses on frequent itemsets in group transaction records. However, it may not directly capture underlying complex patterns and structures, and its effectiveness in identifying hidden malicious groups may be limited. 3. Methods for learning representations: Deep learning and neural network methods typically require a large amount of data and computational resources. Moreover, optimizing and tuning the parameters during the training process is usually quite challenging.

[0039] Based on this, this application proposes a PageRank-based method for detecting malicious groups. First, a directed graph of the group transaction network and a transition matrix are constructed based on the transaction relationships between users within the group. Then, the PageRank vector of this group transition matrix is ​​calculated using the PowerMethod method. Subsequently, the PageRank feature vector is combined with a fully connected neural network to predict and evaluate the group's maliciousness. The method described in this application can detect suspicious transaction behavior within social groups, enabling the control of suspicious group behavior and reducing the risk of fraud and losses for the platform and users.

[0040] The malicious group mining method based on PageRank proposed in this application has the following advantages: (1) Using groups as units, obtain training samples of normal groups and malicious groups.

[0041] (2) For each group, construct a directed graph of the group user transaction network based on the transaction relationship within the group.

[0042] (3) Based on the amount distribution of a user's payment group transfer within the group, construct the transfer matrix corresponding to the transaction network of that group.

[0043] (4) For each group, calculate the PageRank value of each user within it based on the constructed transaction network and transition matrix.

[0044] (5) For each group, merge the PageRank values ​​of each user into a feature vector, which is regarded as the PageRank feature vector of the group.

[0045] (6) Based on the feature vectors obtained above, input them into a fully connected neural network to train a binary classification model.

[0046] (7) For the group to be predicted, construct its corresponding transaction network graph and transition matrix, learn the PageRank value and input it into the trained neural network for prediction, and output whether the group belongs to a malicious group.

[0047] Furthermore, since normal groups and malicious groups (abnormal groups) typically have different transaction patterns, especially since abnormal groups often have a large number of users acting as transaction recipients, resulting in higher PageRank values, this application's embodiments primarily involve constructing a group's PageRank feature vector based on user transaction relationships within the group, and using a neural network to predict whether the target group belongs to the target type of group, such as an abnormal group. For identified abnormal groups, further identification can be performed on the group and users within it with abnormally high PageRank values ​​to reduce the risk of fraud and losses for the platform and users.

[0048] The solutions provided in this application involve technologies such as machine learning in artificial intelligence, and are specifically illustrated through the following embodiments.

[0049] Artificial intelligence (AI) is the theory, methods, technology, and application systems that use digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to achieve optimal results. In other words, AI is a comprehensive technology within computer science that attempts to understand the essence of intelligence and produce a new kind of intelligent machine that can react in a way similar to human intelligence. AI studies the design principles and implementation methods of various intelligent machines, enabling them to possess the functions of perception, reasoning, and decision-making.

[0050] Artificial intelligence (AI) is a comprehensive discipline encompassing a wide range of fields, including both hardware and software technologies. Fundamental AI technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing, operating / interactive systems, and mechatronics. AI software technologies primarily include computer vision, speech processing, natural language processing, and machine learning / deep learning.

[0051] Machine learning (ML) is a multidisciplinary field involving probability theory, statistics, approximation theory, convex analysis, and algorithm complexity theory. It specifically studies how computers can simulate or implement human learning behavior to acquire new knowledge or skills and reorganize existing knowledge structures to continuously improve their performance. Machine learning is the core of artificial intelligence and the fundamental way to endow computers with intelligence; its applications span all areas of artificial intelligence. Machine learning and deep learning typically include techniques such as artificial neural networks, belief networks, reinforcement learning, transfer learning, inductive learning, and learn-by-doing.

[0052] With the research and advancement of artificial intelligence (AI) technology, AI is being studied and applied in various fields, such as smart homes, smart wearable devices, virtual assistants, smart speakers, smart marketing, autonomous driving, drones, robots, smart healthcare, and smart customer service. It is believed that with the development of technology, AI will be applied in more fields and play an increasingly important role.

[0053] The social network group detection method provided in this application can be applied to social network group detection scenarios. For example, in instant messaging applications, considering user privacy and property security, the social network group detection method provided in this application is used to detect whether a target group is a target type of group, such as a sensitive group or a malicious group. If it is determined that the target group is a sensitive group or a malicious group, the target group is blocked or otherwise processed to avoid causing losses to users.

[0054] The method provided in this application is applied to an electronic device, which is a terminal or a server. Optionally, the terminal can be a portable, pocket-sized, or handheld device, such as a mobile phone, computer, or tablet computer. The server is an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDN), and big data and artificial intelligence platforms.

[0055] It should be noted that, in the optional embodiments of this application, the related data such as object information (transaction data volume of each transaction between objects in the target group, chat data between objects in the target group, etc.) require the permission or consent of the object when the embodiments of this application are applied to specific products or technologies. Furthermore, the collection, use, and processing of related data must comply with the relevant laws, regulations, and standards of the relevant countries and regions. In other words, if the embodiments of this application involve data related to objects, it needs to be obtained with the authorization and consent of the object, the authorization and consent of relevant departments, and in accordance with the relevant laws, regulations, and standards of the country and region. If the embodiments involve personal information, the acquisition of all personal information requires the consent of the individual. If sensitive information is involved, the separate consent of the information subject is required. The embodiments also need to be implemented with the authorization and consent of the object.

[0056] Furthermore, it should be noted that the user or object mentioned in the embodiments of this application may refer to account information representing users or objects within a group. To better understand the design concept of this application, the following examples illustrate application scenarios in the embodiments of this application. Please refer to... Figure 1 This paper provides an application scenario for a method of detecting social network groups, which may include a terminal device 110 and a server 120; the terminal device 110 and the server 120 can communicate through a network, wherein: A target client can be installed on terminal device 110. After the target management object logs into the target client, it can trigger the social network group detection method shown in this application embodiment (sending a trigger command to server 120). After receiving the trigger command, server 120 can determine the target group to be detected. As an embodiment, the social network group detection method provided in this application embodiment can be deployed on server 120 for execution. Server 120 obtains the target object group, acquires the transaction data volume of each transaction between each object in the target group, and for each transaction initiator in each object, calculates the first total transaction data volume between the transaction initiator and each transaction receiver corresponding to the transaction initiator based on the transaction data volume of each transaction initiated by the transaction initiator. Calculate the second total transaction data volume of transactions initiated by the transaction initiator. Then, group the transaction initiator and the transaction receiver among the objects into one object group. For each object group, calculate the ratio between the first total transaction data volume of the object group and the second total transaction volume of the transaction initiator in the object group. Then, use the ratio of each object group as the degree of influence of the transaction initiator on the transaction receiver in the object group. For the ratio of each object group in the target group, calculate the transaction importance of each object in the target group in the target group. Then, based on the transaction importance of each object in the target group, perform group classification processing on the target group through the trained classification model to obtain the probability that the target group belongs to the target type group.

[0057] As one embodiment, the social network group detection method provided in this application can also be deployed on terminal device 110 for execution. The execution process can be found in the execution process deployed on server 120, and will not be repeated here. In another possible implementation, if the social network group detection method shown in this application is executed on server 120, the method can be executed without receiving instructions triggered by terminal device 110. This allows the probability that the target group belongs to the target type group to be transmitted to terminal device 110, or alternatively, the probability of whether the target group belongs to the target type group to be transmitted to terminal device 110.

[0058] The terminal device 110 in this embodiment of the application (such as, but not limited to, including) Figure 1 The 110-1 or 110-2 shown in the diagram can be a mobile terminal, a fixed terminal, a portable terminal, or a vehicle-mounted terminal, such as a mobile phone, site, unit, device, multimedia computer, multimedia tablet, Internet node, communicator, desktop computer, laptop computer, notebook computer, netbook computer, tablet computer, personal communication system (PCS) device, personal navigation device, personal digital assistant (PDA), audio / video player, digital camera or camcorder, positioning device, television receiver, radio broadcast receiver, e-book device, gaming device, or any combination thereof, including accessories and peripherals of such devices or any combination thereof.

[0059] The server 120 in this embodiment can be an independent physical server, a server cluster or distributed system composed of multiple physical servers, or multiple cloud servers (such as, but not limited to, servers 120-1, 120-2 or 120-3 shown in the figure) that provide basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms. The functions of the server 120 can be implemented by one or more cloud servers, or by one or more cloud server clusters, etc.

[0060] To facilitate a better understanding and explanation of the methods provided in the embodiments of this application, the optional implementation methods of the methods provided in this application will first be described below with reference to a specific scenario embodiment. In the system architecture shown above, the method can be the terminal device or server shown in the embodiments above. This application embodiment provides a method for detecting social network groups, such as... Figure 2 As shown, the method may include: Step S201: Obtain the transaction data volume of each transaction between each object in the target group.

[0061] In this embodiment of the application, the transaction data volume of each transaction between each object in the target group within a preset time period can be obtained.

[0062] Specifically, in this embodiment of the application, the transaction data volume of each transaction between each object in the target group can be obtained from the backend server corresponding to the social group, or the transaction data volume of each transaction between each object in the target group can be obtained from the transaction records or chat data corresponding to the target group.

[0063] Step S202: For each transaction initiator in each object, based on the transaction data volume of each transaction initiated by the transaction initiator, calculate the first total transaction data volume between the transaction initiator and each transaction receiver corresponding to the transaction initiator, and calculate the second total transaction data volume of the transactions initiated by the transaction initiator.

[0064] For example, the target social group includes four objects: object A, object B, object C, and object D. Taking object A as an example, for object A, calculate the first total transaction data volume between object A and object A (total transaction data volume a1), the first total transaction data volume between object A and object B (total transaction data volume a2), the first total transaction data volume between object A and object C (total transaction data volume a3), and the first total transaction data volume between object A and object D (total transaction data volume a4). Then calculate the second total transaction data volume (total transaction data A) initiated by object A to each object (A, B, C, and D). That is, the second total transaction data volume A of the transactions initiated by object A is the sum of the total transaction data volume a1, total transaction data volume a2, total transaction data volume a3, and total transaction data volume a4. In other words, A = a1 + a2 + a3 + a4. Taking object B as an example, for object B, calculate the first total transaction data volume between object B and object A (total transaction data volume b1), the first total transaction data volume between object B and object B (total transaction data volume b2), the first total transaction data volume between object B and object C (total transaction data volume b3), and the first total transaction data volume between object B and object D (total transaction data volume b4). Then calculate the second total transaction data volume (total transaction data B) initiated by object B to each object (A, B, C, and D). That is, the second total transaction data volume B of the transactions initiated by object B is the sum of the total transaction data volume b1, total transaction data volume b2, total transaction data volume b3, and total transaction data volume b4. In other words, B = b1 + b2 + b3 + b4.

[0065] It should be noted that the first total transaction data volume between a transaction initiator and a corresponding transaction receiver is the sum of the transaction data volumes between the transaction initiator and the transaction receiver within a preset time period.

[0066] If object A is the transaction initiator and object B is the transaction receiver, and within a preset time period, there are two transaction data volumes: a21 and a22. Then, the first total transaction data volume a2 between object A as the transaction initiator and object B as the transaction receiver is the sum of transaction data volumes a21 and a22, that is, a2 = a21 + a22.

[0067] Step S203: Group the transaction initiator and the transaction receiver in each object into an object group. For each object group, calculate the ratio between the first total transaction data volume corresponding to the object group and the second total transaction volume corresponding to the transaction initiator in the object group.

[0068] Continuing with the previous example, consider objects A and B. Object A, as the transaction initiator, and object B, as the transaction recipient, belong to one object group, denoted as object group 1. Object A, as the transaction recipient, and object B, as the transaction initiator, belong to another object group, denoted as object group 2.

[0069] Specifically, for object group 1, the first total transaction data volume corresponding to object group 1 (object A as the transaction initiator and object B as the transaction receiver) is calculated as a1, and the second total transaction data volume corresponding to object group A is calculated as A. The ratio between the first total transaction data volume a1 corresponding to object group 1 and the second total transaction data volume A corresponding to object A as the transaction initiator in object group 1 is calculated, which is a1 / A. For object group 2, the first total transaction data volume corresponding to object group 2 (object B as the transaction initiator and object A as the transaction receiver) is calculated as b1, and the ratio between the first total transaction data volume b1 corresponding to object group 2 (object B as the transaction initiator) and the second total transaction data volume B corresponding to object B as the transaction initiator in object group 2 is calculated, which is b1 / B.

[0070] Step S204: Take the ratio corresponding to each object group as the degree of influence of the transaction initiator on the transaction recipient in that object group. For the ratio corresponding to each object group in the target group, calculate the transaction importance of each object in the target group.

[0071] In the embodiments of this application, after obtaining the ratio corresponding to each object group, a matrix is ​​constructed based on the ratio corresponding to each object group, and then the transaction degree of each object in the target group is calculated based on the matrix. The transaction importance of each object in the target group can also be represented by the matrix.

[0072] Step S205: Based on the transaction importance of each object in the target group, the target group is classified using the trained classification model to obtain the probability that the target group belongs to the target type group.

[0073] In the embodiments of this application, the trained classification model can be a trained fully connected neural network or other types of trained machine learning models, such as XGB (eXtreme Gradient Boosting).

[0074] Specifically, the target type can be a preset type, such as an abnormal group type.

[0075] Furthermore, after obtaining the transaction importance of each object in the target group, the transaction importance of each object in the target group is input into the trained classification model, and the probability of the target group belonging to the target type group is output. Further, after determining the probability that the target group belongs to the target type group, the relationship between this probability and a preset probability threshold can be used to determine whether the target group belongs to the target type group.

[0076] For example, if the preset probability threshold is 0.9, and the probability that the target group belongs to the target type group is 0.95, then the target group is determined to belong to the target type group; if the probability that the target group belongs to the target type group is 0.6, then the target group is determined not to belong to the target type group.

[0077] This application provides a method for detecting social network groups. In this application, by acquiring the transaction data volume of each transaction between each object in the target group, and for each transaction initiator in each object, the first total transaction data volume between the transaction initiator and each corresponding transaction recipient of the transaction initiator and the second total transaction data volume of the transaction initiated by the transaction initiator can be calculated. Since a transaction initiator and a transaction recipient in each object are considered as an object group, the ratio between the first total transaction data volume corresponding to each object group and the second total transaction volume corresponding to the transaction initiator in the object group can be obtained based on the first total transaction data volume and the second total transaction data volume. This ratio is used as the degree of influence of the transaction initiator in the object group on the transaction recipient. Based on this ratio, the transaction importance of each object in the target group in the target group can be calculated. Based on the transaction importance of each object in the target group in the target group, the probability of the target group belonging to the target type group can be obtained through a trained classification model, thereby determining whether the target group belongs to the target type group.

[0078] Specifically, the method for obtaining the transaction data volume of each transaction between each object in the target group in step S201 is detailed in the above embodiment. For each transaction initiator in each object, based on the transaction data volume of each transaction initiated by the transaction initiator, the first total transaction data volume between the transaction initiator and each transaction receiver corresponding to the transaction initiator is calculated, and the second total transaction data volume of the transactions initiated by the transaction initiator is calculated. In the embodiments of this application, the second total transaction data volume corresponding to each transaction initiator in each object is obtained in the following way: each object in the target group is treated as a node, and an edge is constructed between the nodes corresponding to the objects with transactions, pointing from the node of the transaction initiator to the transaction receiver. The first total transaction data volume between the objects is used as the weight to construct a directed graph of transaction relationships corresponding to the target group; for each source node of the directed graph of transaction relationships, the second total transaction data volume of the transaction initiator corresponding to the source node is obtained by calculating the sum of the weights of each edge sent by the source node.

[0079] For example, the target group includes: object A, object B, object C, and object D. The transaction relationships between the four objects in the target group are shown in Table 1. Table 1

[0080] In other words, the target group includes: object A, object B, object C, and object D. Object A, as the transaction initiator, has transaction relationships with objects B, C, and D respectively. The total data volume of the first transaction between object A and object B is 1000, the first transaction between object A and object C is 2000, and the first transaction between object A and object D is 5000. Object B, as the transaction initiator, has transaction relationships with both object A and object D. The total data volume of the first transaction between object B and object D is 2000, and the first transaction between object B and object A is 100. Object C, as the transaction initiator, has a transaction relationship with object D, and the total data volume of the first transaction between object C and object D is 10000. Object D, as the transaction initiator, has a transaction relationship with object C, and the total data volume of the first transaction between object D and object C is 5000. Using objects A, B, C, and D as nodes, a directed graph of transaction relationships is constructed, as follows: Figure 3As shown, the direction of the directed graph of transactions is from the transaction initiator to the transaction receiver. Taking object A as an example, the nodes corresponding to object A point to objects B, C, and D respectively, resulting in the edges from object A to object B, object A to object C, and object A to object D. The weight of the edge from object A to object B is equal to the first total transaction data volume between object A and object B; the weight of the edge from object A to object C is equal to the first total transaction data volume between object A and object D; and so on. Furthermore, based on the sum of the weights of the edges from object A to object B (which is equal to the first total transaction data volume between object A and object B), the weights of the edges from object A to object C (which is equal to the first total transaction data volume between object A and object C), and the weights of the edges from object A to object C (which is equal to the first total transaction data volume between object A and object D), the second total transaction data volume corresponding to object A is obtained. It should be noted that... Figure 3 The weights are not shown in the text.

[0081] Furthermore, in the above embodiments, after obtaining the first total transaction data volume between the transaction initiator and each transaction recipient corresponding to the transaction initiator, and the second total transaction data volume of the transactions initiated by the transaction initiator, since the second total transaction data volume of a transaction initiated by a certain party may be 0, that is, at this time, the transaction initiator may not have initiated any transactions, in order to avoid errors in calculating the ratio between the first total transaction data volume corresponding to the object group and the second total transaction volume corresponding to the transaction initiator in the object, before calculating the ratio between the first total transaction data volume corresponding to the object group and the second total transaction volume corresponding to the transaction initiator in the object group, the method further includes: determining that the second total transaction volume corresponding to the transaction initiator in the object group is greater than or equal to a preset transaction volume threshold. In other words, for each object group, before calculating the ratio between the first total transaction volume corresponding to the object group and the second total transaction volume corresponding to the transaction initiator in the object group, the relationship between the transaction initiator in the object group and the preset transaction volume threshold is determined. If the second total transaction volume corresponding to the transaction initiator in the object group is greater than or equal to the preset transaction volume threshold, then the ratio between the first total transaction data volume corresponding to the object group and the second total transaction volume corresponding to the transaction initiator in the object group is determined.

[0082] Continuing with the previous example, the ratio between the first total transaction data volume (b1) corresponding to object group 2 (object B as the transaction initiator and object A as the transaction recipient) and the second total transaction data volume (B) corresponding to object B as the transaction initiator in object group 2 is b1 / B.

[0083] Furthermore, if the total transaction volume corresponding to the transaction initiator in the object group is less than a preset transaction volume threshold, then the ratio between the first total transaction volume corresponding to the object group and the preset transaction volume threshold is calculated, and this ratio is used as the ratio between the first total transaction data volume corresponding to the object group and the second total transaction volume corresponding to the transaction initiator in the object group. For example, the preset transaction volume threshold can be 1, then the ratio between the first total transaction volume corresponding to the object group and the preset transaction volume threshold is the first total transaction data volume / 1. In one possible case, the first total transaction data volume is 0, that is, the ratio between the first total transaction volume corresponding to the object group and the preset transaction volume threshold is 0.

[0084] Specifically, based on the first number of objects in the target group and the ratio corresponding to each object group, a first matrix is ​​constructed in which the number of rows and columns are equal to the first number, and the element value at any position is equal to the ratio corresponding to the object group at that position.

[0085] In this first matrix, both the number of rows and columns represent the number of objects contained in the target group, and the element at any position in the first matrix represents the ratio corresponding to the object group at that position. In this embodiment, the first matrix can be the transition matrix M of the transaction network of the target group, which is obtained based on the distribution of transactions in a certain group within the target group; Specifically, given a transaction network containing N objects (nodes), its transition matrix is ​​an N-order matrix M, as shown below: ; Where N represents the number of objects contained in the target group, and m is the element in the i-th row and j-th column. ij This is the ratio between the transaction volume from node j to node i and the total transaction volume received by node j as the transaction initiator. Note that the transaction volume from node j to node i may be zero at some point; therefore, m is calculated using the following formula. ij .

[0086] ; The `max` in the denominator is used to avoid a denominator of 0, which indicates that the node has no transactions. Here, `m`... ij The value represents the ratio between the transaction volume from node j to node i and the total transaction volume received by node j as the transaction initiator. In the transition matrix, it represents the probability of node j transitioning to node i. The larger the value for node i, the greater the proportion of the object corresponding to that node in the transaction network as a transaction receiver (receiver), and the more active the receiving activity.

[0087] For example, for the directed graph of the transaction network shown above ( Figure 3 Taking as an example, its corresponding transition matrix can be expressed as: .

[0088] For the embodiments of this application, the transaction data volume of a single transaction includes at least one of the transaction resource volume and the number of transactions; that is, in the above embodiments, for each object group, the ratio between the first total transaction data volume corresponding to the object group and the second total transaction volume corresponding to the transaction initiator in the object group can be a first ratio between the first transaction resource volume corresponding to the object group and the second transaction resource volume corresponding to the transaction initiator in the object group, or it can be a second ratio between the first number of transactions corresponding to the object group and the second number of transactions corresponding to the transaction initiator in the object group, or it can be the ratio between the first total transaction data volume corresponding to the object group and the second total transaction volume corresponding to the transaction initiator in the object group obtained based on the first ratio and the second ratio.

[0089] Specifically, if the transaction data volume of a single transaction includes: transaction resource volume and transaction count, then in step S204, for each object group, the ratio between the first total transaction data volume corresponding to the object group and the second total transaction volume corresponding to the transaction initiator in the object group can be calculated. This can specifically include: calculating the first ratio between the first total transaction resource volume corresponding to the object group and the second total transaction resource volume corresponding to the transaction initiator in the object group; calculating the second ratio between the first total transaction count corresponding to the object group and the second total transaction count corresponding to the transaction initiator in the object group; obtaining the target weight information corresponding to the first ratio and the second ratio respectively; and calculating the ratio between the first total transaction data volume corresponding to the object group and the second total transaction volume corresponding to the transaction initiator in the object group based on the first ratio, the second ratio, and their respective weight information.

[0090] Specifically, in this embodiment of the application, for a group of objects, after obtaining the first ratio and the second ratio corresponding to the group of objects, the ratio between the first total transaction data volume corresponding to the group of objects and the second total transaction volume corresponding to the transaction initiator in the group of objects is calculated based on the first ratio, the second ratio and their respective weight information.

[0091] It should be noted that the target weight information corresponding to the first ratio and the second ratio can be obtained before obtaining the first ratio and the second ratio respectively, or the target weight information corresponding to the first ratio and the second ratio can be obtained after obtaining the first ratio and the second ratio respectively, or the first ratio and the second ratio can be obtained at the same time as obtaining the target weight information corresponding to the first ratio and the second ratio respectively.

[0092] Specifically, for each object group, the target weight information corresponding to the first ratio and the second ratio is obtained. This may include: obtaining the first weight information corresponding to the first ratio and the second ratio in the object group; obtaining the transaction habit characteristics of the transaction initiator in the object group, and the time interval between two adjacent transactions in the object group; adjusting the first weight information based on the transaction habit characteristics and the time interval between two adjacent transactions to obtain the target weight information corresponding to the first ratio and the second ratio.

[0093] The first weight information is obtained by weighting the number of objects in the target group, the first ratio, and the second ratio using a weighting model. In this embodiment, the number of objects in the target group, the first ratio, and the second ratio are input into a weighting model for calculation to obtain the first weight information corresponding to each of the first and second ratios. In one possible case, the first weight information corresponding to each of the first and second ratios can be used as the target weight information corresponding to each of the first and second ratios. To further improve the accuracy of the target weight information and the accuracy of target group type detection, after obtaining the first weight information, the transaction habit characteristics of the transaction initiator in the object group and the time interval between two adjacent transactions in the object group can be obtained. Then, based on the transaction habit characteristics and the time interval between two adjacent transactions, the first weight information is adjusted to obtain the target weight information corresponding to each of the first and second ratios.

[0094] Among them, the transaction initiator's transaction habit characteristics can characterize the transaction resource amount that the transaction initiator is accustomed to using in a single transaction, or other transaction habits, which may affect the weight information between the first ratio and the second ratio. Of course, if the time interval between two adjacent transactions in the object group is relatively short, the transaction resource amount in a single transaction may be small, or a single transaction may be obtained through the transaction resource amount of multiple transactions, which may also affect the weight information between the first ratio and the second ratio. Therefore, after obtaining the time interval between two adjacent transactions in the object group, and based on the transaction habit characteristics and the time interval between two adjacent transactions, a specific algorithm can be used, for example, a weight adjustment model, to adjust the first weight information corresponding to the first ratio and the second ratio respectively, so as to obtain the target weight information corresponding to the first ratio and the second ratio respectively.

[0095] Specifically, after obtaining the ratios corresponding to each object group in the target group, the ratio corresponding to each object group can be the ratio corresponding to that object group obtained based on the first ratio, the second ratio, and their respective target weight information, or it can be either the first ratio or the second ratio. To improve the accuracy of calculating the transaction importance of each object in the target group, for the ratios corresponding to each object group in the target group, the transaction importance of each object in the target group in the target group is calculated. Specifically, this can include: obtaining the directed graph of target transaction reasons corresponding to the target transaction group; taking one transaction initiator and one transaction receiver in each object as an object group; for each object group in the target group, determining the weight of the edge corresponding to that object group from the directed graph of target transaction reasons; adjusting the ratio between the first total transaction data volume corresponding to that object group and the second total transaction volume corresponding to the transaction initiator in that object group based on the weight of the edge corresponding to that object group; and calculating the transaction importance of each object in the target group based on the adjusted ratios corresponding to each object group.

[0096] Specifically, the method for obtaining the directed graph of the target transaction reasons corresponding to the target transaction group can include: steps Sa, Sb, Sc, Sd, and Se, where, Step Sa: Obtain the relationship diagram between objects in the target group. Specifically, in this embodiment, chat data of a target group within a preset time period is obtained. Based on this chat data, the intimacy level between objects can be analyzed. Specifically, chat data between any two objects can be extracted from the chat data within the preset time period, and keywords can be extracted from the chat data to determine the intimacy relationship between the two objects. Alternatively, the chat data can be input into an intimacy analysis model to obtain the intimacy level between any two objects. After determining the intimacy level between objects, a relationship graph is constructed, where each node in the relationship graph represents an object in the target group, and the weight of the edge between any two objects represents the intimacy level between the two objects. The weight of the edge between any two objects is directly proportional to the intimacy level between the two objects.

[0097] Step Sb: Obtain chat data between various objects in the target group, and extract the context chat data corresponding to the transactions between each object from the chat data.

[0098] In one possible implementation, instead of obtaining the chat data between the objects in the target group, you can directly obtain the context chat data corresponding to the transactions between the objects.

[0099] Step Sc: Construct a directed graph of the first transaction reason based on the context chat data corresponding to each object when a transaction occurs.

[0100] In the embodiments of this application, after obtaining the context chat data corresponding to the transactions of each object, the reason for each transaction can be determined from the context chat data to determine the reliability of each transaction relationship. Specifically, for each transaction, data that can characterize the reason for the transaction or chat data that characterizes the intention to trade can be filtered from the context chat data corresponding to the transaction to determine the reliability of each transaction relationship; or, the context chat data at the time of each transaction can be directly encoded to obtain its corresponding vector, and the reliability of each transaction relationship can be determined through the transaction reliability reason analysis model.

[0101] In the first transaction reason, a node in the directed graph represents an object. There is an edge between the nodes corresponding to the two objects that have a transaction. The direction of the edge is from the node of the transaction initiator to the node of the transaction receiver. The weight of the edge is used to represent the reliability of the transaction relationship, and the weight of the edge is directly proportional to the reliability of the transaction relationship.

[0102] It should be noted that step Sa can be executed before step Sb-Step Sc, after step Sc, between step Sb and step Sc, or simultaneously with either step Sb or step Sc.

[0103] Step Sd: Based on the relationship graph between objects in the target group, adjust the weights of each edge in the first directed graph of transaction reasons to obtain the second directed graph of transaction reasons.

[0104] To further improve the accuracy of the importance of each object in the target group in terms of transactions within the target group, the reliability of the directed graph of the determined transaction causes can be improved. Since the weight of each edge in the first directed graph of transaction causes represents the reliability of the transaction between the two objects connected by that edge, the weights of each edge in the first directed graph of transaction causes can be adjusted by using the relationship graph between the objects in the target group to obtain a more accurate second directed graph of transaction causes.

[0105] Step Se: Determine at least one of the first and second directed graphs of transaction reasons as the target transaction reason directed graph corresponding to the target transaction group.

[0106] Specifically, when determining the directed graph of the target transaction reasons corresponding to the target transaction group, the first directed graph of the transaction reasons can be directly determined as the directed graph of the target transaction reasons corresponding to the target transaction group, or the second directed graph of the transaction reasons can be directly determined as the directed graph of the target transaction reasons corresponding to the target transaction group, or the first directed graph of the transaction reasons and the second directed graph of the transaction reasons can be merged to obtain the directed graph of the target transaction reasons corresponding to the target transaction group.

[0107] As can be seen from the above embodiments, in the directed graph of the target transaction cause, a node represents an object, and there is an edge between the nodes corresponding to the two objects that have a transaction. The direction of the edge is from the node of the transaction initiator to the node of the transaction receiver. The weight of the edge is used to represent the reliability of the transaction relationship, and the weight of the edge is directly proportional to the reliability of the transaction relationship. In addition, the weight of each edge in the directed graph of the transaction cause is inversely proportional to the ratio of the group of objects corresponding to that edge.

[0108] For each object group in the target group, the weight of the edge corresponding to that object group is determined from the directed graph of the target transaction causes. Based on the weight of the edge corresponding to that object group, the ratio between the first total transaction data volume corresponding to that object group and the second total transaction volume corresponding to the transaction initiator in that object group is adjusted. Based on the adjusted ratios corresponding to each object group, the transaction importance of each object in the target group is calculated. In this embodiment, for each object group in the target group, based on the weight of the edge corresponding to that object group, i.e., the reliability of the transaction occurring in that object group, the ratio of the first total transaction data volume and the second total transaction data volume in that object group is adjusted to obtain the adjusted ratios corresponding to each object group. Then, based on the adjusted ratios corresponding to each object group, the transaction importance of each object in the target group is calculated.

[0109] Specifically, for the ratios corresponding to each object group in the target group, the transaction importance of each object in the target group can be calculated. This can include: constructing a first matrix based on the first number of objects in the target group and the ratios corresponding to each object group, where the number of rows and columns is equal to the first number, and the element value at any position is equal to the ratio of the object group corresponding to that position; wherein the number of rows and columns of the first matrix are the number of objects contained in the target group, and the element at any position in the first matrix is ​​the ratio of the object group represented at that position; constructing an initial feature vector, and calculating the feature vector of the first matrix based on the initial feature vector and the first matrix, wherein the feature vector of the first matrix is ​​used to represent the transaction importance of each object in the target group.

[0110] For the embodiments of this application, the method of constructing the first matrix can be detailed in the above embodiments. After constructing the first matrix, which can also be called the transition matrix, the eigenvectors of the first matrix are calculated to obtain the transaction importance of each object in the target group in the target group.

[0111] Specifically, after constructing the transition matrix, the next step is to calculate the PageRank value of each node, or the PageRank vector of the transition matrix.

[0112] The PageRank vector of the transition matrix satisfies the following relationship: the PageRank vector is an eigenvector of the transition matrix. ; Where M is the transition matrix and P is the eigenvector of the transition matrix.

[0113] Specifically, the PowerMethod can be used to calculate the PageRank vector. The specific calculation formula is as follows:

[0114] M ; Where i = 0, 1, 2…; Here, R0 is a random vector, which is the initial feature vector shown above.

[0115] Based on the above embodiments, after obtaining the PageRank vector of the transition matrix, which is the transaction importance of each object in the target group, the target group is classified by the trained classification model to obtain the probability that the target group belongs to the target type group.

[0116] Specifically, to avoid different feature vectors due to the order of object arrangement, the PageRank values ​​of each object need to be sorted in descending order. Specifically, based on the transaction importance of each object in the target group, the trained classification model is used to classify the target group, obtaining the probability that the target group belongs to the target type group. This can include: sorting the feature vectors of the first matrix according to a preset order to obtain sorted feature vectors; and using the sorted feature vectors, performing group classification through the trained classification model to obtain the probability that the target group belongs to the target type group.

[0117] Specifically, sorting the eigenvectors of the first matrix according to a preset order to obtain sorted eigenvectors can include: reordering each element of the eigenvectors of the first matrix according to its size to obtain sorted eigenvectors.

[0118] For example, a feature vector dimension K is pre-defined. The top K values ​​from the calculated PageRank vector, sorted from largest to smallest, are combined into a single feature vector. If there are fewer than K values, zeros are added to the end of the vector.

[0119] Furthermore, after obtaining the sorted feature vectors, group classification is performed based on the sorted feature vectors and the trained classification model to obtain the probability that the target group belongs to the target type group.

[0120] As shown above, the classification model can be a fully connected neural network. Figure 4 This is a schematic diagram of a two-layer fully connected neural network. The input is the PageRank feature vector (the feature vector after sorting). After processing by the intermediate layer, the output is the corresponding class probability vector.

[0121] Furthermore, if the probability that the target group belongs to the target type group is greater than the preset probability threshold, that is, the target group belongs to the target type group, which may belong to some abnormal type groups. In this case, there may be some abnormal accounts or objects in these abnormal groups. Therefore, after the probability that the target group belongs to the target type group is greater than the preset probability threshold, the objects in the target group whose transaction importance is greater than the preset importance threshold are identified as target objects.

[0122] The preset importance threshold can be either pre-stored in an electronic device by the administrator or entered by the administrator through a terminal device.

[0123] Furthermore, the probability of a target group belonging to a target type group is obtained by performing group classification processing on the trained classification model.

[0124] Specifically, the trained classification model is obtained by training the initial model in the following ways (steps S1, S2, S3, and S4), as follows: Figure 5 As shown: Step S1: Obtain multiple training samples.

[0125] In this embodiment of the application, training samples of normal and abnormal groups are obtained for training.

[0126] Each training sample includes: the transaction data volume between objects in a sample group, and the label corresponding to that sample group. The label of a sample group is used to characterize whether the sample group belongs to the target type of group. A large number of group transactions are based on normal user transactions, such as family groups, work groups, etc. For this part of the groups, we randomly sample a portion as our normal group sample, with a label of 0. We also have some abnormal groups, such as fraudulent order groups, etc. We treat all these malicious groups as our abnormal group sample, with a label of 1.

[0127] Step S2: For each sample group, obtain the transaction data volume of each transaction between each object in the sample group.

[0128] Specifically, for a certain sample group, all group transactions within a certain time window (such as half a year, 1 year, etc.) are obtained. Specifically for each sample group, the method of obtaining the transaction data volume of each transaction between each object in the sample group is similar to the method of obtaining the transaction data volume of each transaction between each object in the current group. For details, please refer to the above embodiment, which will not be repeated here.

[0129] Step S3: For each transaction initiator in each object, based on the transaction data volume of each transaction initiated by the transaction initiator, calculate the first total transaction data volume between the transaction initiator and each transaction receiver corresponding to the transaction initiator, and calculate the second total transaction data volume of the transactions initiated by the transaction initiator.

[0130] Specifically, for each transaction initiator among all objects, the method of calculating the first total transaction data volume between the transaction initiator and each corresponding transaction recipient based on the transaction data volume of each transaction initiated by the transaction initiator, and the method of calculating the second total transaction data volume of the transactions initiated by the transaction initiator, is the same as the method of calculating the first total transaction data volume and the second total transaction data volume for objects in the target group. For details, please refer to the above embodiments, and will not be repeated here.

[0131] Step S4: Group the transaction initiator and the transaction receiver in each object into an object group. For each object group, calculate the ratio between the first total transaction data volume corresponding to the sample object group and the second total transaction volume corresponding to the transaction initiator in the object group.

[0132] Specifically, for each object group in the sample group, the method of calculating the ratio between the first total transaction data volume corresponding to the sample object group and the second total transaction volume corresponding to the transaction initiator in the object group is the same as the method of calculating the ratio between the first total transaction data volume corresponding to the object group and the second total transaction volume corresponding to the transaction initiator in the object group for each object group in the target group. For details, please refer to the above embodiments, and they will not be repeated here.

[0133] Step S5: Take the ratio corresponding to each sample object group as the degree of influence of the transaction initiator on the transaction recipient in that sample object group. For the ratio corresponding to each object group in the sample group, calculate the transaction importance of each object in the sample group in the sample group.

[0134] In the embodiments of this application, the method for calculating the transaction importance of each object in the sample group is the same as the method for calculating the transaction importance of each object in the target group. For details, please refer to the above embodiments, and will not be repeated here.

[0135] Step S6: Based on the importance of each sample object in the transaction within each sample group and the label corresponding to that sample group, train the initial model to obtain the trained classification model.

[0136] Specifically, after obtaining the PageRank feature vectors of each sample group, further classification model training is performed. Figure 4 This is a schematic diagram of a two-layer fully connected neural network. The input is the PageRank feature vector, and the output is the corresponding class probability vector. The activation function used is the sigmoid function.

[0137] Specifically, in this embodiment of the application, the group PageRank vector is input into the network model for classification training.

[0138] The calculation formula for the model is as follows: ; Where w1, w2, b1, and b2 are learnable model parameters.

[0139] The corresponding loss function is: ; Where y is the label corresponding to the sample, and the parameter training process adopts gradient descent.

[0140] Once the model is trained, its performance is evaluated to determine whether further training is needed. Common evaluation metrics include IV, AUC (Area Under the Curve), and KS (Kolmogorov-Smirnov). If the model performs as expected, it can be deployed for actual prediction tasks; if it does not perform as expected, it needs to be retrained.

[0141] Once the above model has been trained, it can be used to assess and predict the malice level of new group samples.

[0142] Furthermore, the method for detecting social network groups as shown in the embodiments of this application will be described in the following specific examples, such as... Figure 6 As shown, training samples are first obtained, and a directed graph of the sample group object transaction network is constructed based on the training samples. Then, a transition matrix is ​​constructed based on the directed graph of the transaction network. Further, PageRank is calculated based on the transition matrix to obtain the PageRank vector corresponding to the sample group. Based on the PageRank vector, the fully connected neural network classification model is trained through the constructed fully connected network. Then, the training effect of the network is evaluated. If the effect does not meet expectations, the model can be retrained. If the effect meets expectations, the transaction data volume of the group to be predicted (target group) is predicted through the model to obtain the prediction result, that is, the probability of whether the target group belongs to the target type group.

[0143] To illustrate the examples in this application in more detail, specifically, such as Figure 7 As shown, where, Step S701: Obtain multiple training samples.

[0144] In this embodiment of the application, training samples of normal and abnormal groups are obtained for training.

[0145] Each training sample includes: the transaction data volume between objects in a sample group, and the label corresponding to that sample group. The label of a sample group is used to characterize whether the sample group belongs to the target type of group. A large number of group transactions are based on normal user transactions, such as family groups, work groups, etc. For this part of the groups, we randomly sample a portion as our normal group sample, with a label of 0. We also have some abnormal groups, such as fraudulent order groups, etc. We treat all these malicious groups as our abnormal group sample, with a label of 1.

[0146] Step S702: For each sample group, obtain the transaction data volume of each transaction between each object in the sample group.

[0147] Specifically, for a certain sample group, all group transactions within a certain time window (such as half a year, 1 year, etc.) are obtained. Specifically for each sample group, the method of obtaining the transaction data volume of each transaction between each object in the sample group is similar to the method of obtaining the transaction data volume of each transaction between each object in the current group. For details, please refer to the above embodiment, which will not be repeated here.

[0148] Step S703: For each transaction initiator in each object, based on the transaction data volume of each transaction initiated by the transaction initiator, calculate the first total transaction data volume between the transaction initiator and each transaction receiver corresponding to the transaction initiator, and calculate the second total transaction data volume of the transactions initiated by the transaction initiator.

[0149] Specifically, for each transaction initiator among all objects, the method of calculating the first total transaction data volume between the transaction initiator and each corresponding transaction recipient based on the transaction data volume of each transaction initiated by the transaction initiator, and the method of calculating the second total transaction data volume of the transactions initiated by the transaction initiator, is the same as the method of calculating the first total transaction data volume and the second total transaction data volume for objects in the target group. For details, please refer to the above embodiments, and will not be repeated here.

[0150] Step S704: Group the transaction initiator and the transaction receiver in each object into an object group. For each object group, calculate the ratio between the first total transaction data volume corresponding to the sample object group and the second total transaction volume corresponding to the transaction initiator in the object group.

[0151] Specifically, for each object group in the sample group, the method of calculating the ratio between the first total transaction data volume corresponding to the sample object group and the second total transaction volume corresponding to the transaction initiator in the object group is the same as the method of calculating the ratio between the first total transaction data volume corresponding to the object group and the second total transaction volume corresponding to the transaction initiator in the object group for each object group in the target group. For details, please refer to the above embodiments, and they will not be repeated here.

[0152] Step S705: Take the ratio corresponding to each sample object group as the degree of influence of the transaction initiator on the transaction recipient in that sample object group. For the ratio corresponding to each object group in the sample group, calculate the transaction importance of each object in the sample group in the sample group.

[0153] In the embodiments of this application, the method for calculating the transaction importance of each object in the sample group is the same as the method for calculating the transaction importance of each object in the target group. For details, please refer to the above embodiments, and will not be repeated here.

[0154] Step S706: Based on the importance of each sample object in the transaction within each sample group and the label corresponding to that sample group, train the initial model to obtain the trained classification model.

[0155] Specifically, after obtaining the PageRank feature vectors of each sample group, further classification model training is performed. Figure 4 This is a schematic diagram of a two-layer fully connected neural network. The input is the PageRank feature vector, and the output is the corresponding class probability vector. The activation function used is the sigmoid function.

[0156] Specifically, in this embodiment of the application, the group PageRank vector is input into the network model for classification training.

[0157] The calculation formula for the model is as follows: ; Where w1, w2, b1, and b2 are learnable model parameters.

[0158] The corresponding loss function is: ; Where y is the label corresponding to the sample, and the parameter training process adopts gradient descent.

[0159] Once the model is trained, its performance is evaluated to determine whether further training is needed. Common evaluation metrics include IV, AUC (Area Under the Curve), and KS (Kolmogorov-Smirnov). If the model performs as expected, it can be deployed for actual prediction tasks; if it does not perform as expected, it needs to be retrained.

[0160] Once the above model has been trained, it can be used to assess and predict the malice level of new group samples.

[0161] Step S707: Obtain the transaction data volume of each transaction between each object in the target group.

[0162] In this embodiment of the application, the transaction data volume of each transaction between each object in the target group within a preset time period can be obtained.

[0163] Specifically, in this embodiment of the application, the transaction data volume of each transaction between each object in the target group can be obtained from the backend server corresponding to the social group, or the transaction data volume of each transaction between each object in the target group can be obtained from the transaction records or chat data corresponding to the target group.

[0164] Step S708: For each transaction initiator in each object, based on the transaction data volume of each transaction initiated by the transaction initiator, calculate the first total transaction data volume between the transaction initiator and each transaction receiver corresponding to the transaction initiator. Treat each object in the target group as a node, and construct an edge connecting the nodes corresponding to the objects with transactions, from the node of the transaction initiator to the transaction receiver. Use the first total transaction data volume between objects as weights to construct a directed graph of transaction relationships corresponding to the target group. For each source node in the directed graph of transaction relationships, calculate the sum of the weights of each edge sent by the source node to obtain the second total transaction data volume of the transaction initiator corresponding to the source node.

[0165] For example, the target social group includes four objects: object A, object B, object C, and object D. Taking object A as an example, for object A, calculate the first total transaction data volume between object A and object A (total transaction data volume a1), the first total transaction data volume between object A and object B (total transaction data volume a2), the first total transaction data volume between object A and object C (total transaction data volume a3), and the first total transaction data volume between object A and object D (total transaction data volume a4). Then calculate the second total transaction data volume (total transaction data A) initiated by object A to each object (A, B, C, and D). That is, the second total transaction data volume A of the transactions initiated by object A is the sum of the total transaction data volume a1, total transaction data volume a2, total transaction data volume a3, and total transaction data volume a4. In other words, A = a1 + a2 + a3 + a4. Taking object B as an example, for object B, calculate the first total transaction data volume between object B and object A (total transaction data volume b1), the first total transaction data volume between object B and object B (total transaction data volume b2), the first total transaction data volume between object B and object C (total transaction data volume b3), and the first total transaction data volume between object B and object D (total transaction data volume b4). Then calculate the second total transaction data volume (total transaction data B) initiated by object B to each object (A, B, C, and D). That is, the second total transaction data volume B of the transactions initiated by object B is the sum of the total transaction data volume b1, total transaction data volume b2, total transaction data volume b3, and total transaction data volume b4. In other words, B = b1 + b2 + b3 + b4.

[0166] It should be noted that the first total transaction data volume between a transaction initiator and a corresponding transaction receiver is the sum of the transaction data volumes between the transaction initiator and the transaction receiver within a preset time period.

[0167] If object A is the transaction initiator and object B is the transaction receiver, and within a preset time period, there are two transaction data volumes: a21 and a22. Then, the first total transaction data volume a2 between object A as the transaction initiator and object B as the transaction receiver is the sum of transaction data volumes a21 and a22, that is, a2 = a21 + a22.

[0168] Furthermore, based on Table 1, a directed graph of transaction relationships is constructed, as shown below. Figure 3 As shown, the direction of the directed graph of transactions is from the transaction initiator to the transaction receiver. Taking object A as an example, the nodes corresponding to object A point to objects B, C, and D respectively, resulting in the edges from object A to object B, object A to object C, and object A to object D. The weight of the edge from object A to object B is equal to the first total transaction data volume between object A and object B; the weight of the edge from object A to object C is equal to the first total transaction data volume between object A and object D; and so on. Furthermore, based on the sum of the weights of the edges from object A to object B (which is equal to the first total transaction data volume between object A and object B), the weights of the edges from object A to object C (which is equal to the first total transaction data volume between object A and object C), and the weights of the edges from object A to object C (which is equal to the first total transaction data volume between object A and object D), the second total transaction data volume corresponding to object A is obtained. It should be noted that... Figure 3 The weights are not shown in the text.

[0169] Step S709: If the total transaction volume corresponding to the transaction initiator in the object group is less than the preset transaction volume threshold, calculate the ratio between the first total transaction volume corresponding to the object group and the preset transaction volume threshold, and use this ratio as the ratio between the first total transaction data volume corresponding to the object group and the second total transaction volume corresponding to the transaction initiator in the object group.

[0170] For example, if the preset transaction volume threshold can be 1, then the ratio between the first total transaction volume corresponding to the object group and the preset transaction volume threshold is the first total transaction data volume / 1. In one possible case, the first total transaction data volume is 0, that is, the ratio between the first total transaction volume corresponding to the object group and the preset transaction volume threshold is 0.

[0171] Step S710: Based on the first number of objects in the target group and the ratio corresponding to each object group, construct a first matrix in which the number of rows and columns are equal to the first number, and the element value at any position is equal to the ratio corresponding to the object group at that position.

[0172] In this first matrix, both the number of rows and columns represent the number of objects contained in the target group, and the element at any position in the first matrix represents the ratio corresponding to the object group at that position. In this embodiment, the first matrix can be the transition matrix M of the transaction network of the target group, which is obtained based on the distribution of transactions in a certain group within the target group; Specifically, given a transaction network containing N objects (nodes), its transition matrix is ​​an N-order matrix M, as shown below: ; Where N represents the number of objects contained in the target group, and m is the element in the i-th row and j-th column. ij This is the ratio between the transaction volume from node j to node i and the total transaction volume received by node j as the transaction initiator. Note that the transaction volume from node j to node i may be zero at some point; therefore, m is calculated using the following formula. ij .

[0173] ; The `max` in the denominator is used to avoid a denominator of 0, which indicates that the node has no transactions. Here, `m`... ij The value represents the ratio between the transaction volume from node j to node i and the total transaction volume received by node j as the transaction initiator. In the transition matrix, it represents the probability of node j transitioning to node i. The larger the value for node i, the greater the proportion of the object corresponding to that node in the transaction network as a transaction receiver (receiver), and the more active the receiving activity.

[0174] For example, for the directed graph of the transaction network shown above ( Figure 3 Taking as an example, its corresponding transition matrix can be expressed as: .

[0175] Step S711: After constructing the first matrix, which can also be called the transition matrix, sort the elements in the transition matrix in descending order, and calculate the eigenvector of the first matrix to obtain the transaction importance of each object in the target group.

[0176] Specifically, after constructing the transition matrix, the next step is to calculate the PageRank value of each node, or the PageRank vector of the transition matrix.

[0177] The PageRank vector of the transition matrix satisfies the following relationship: the PageRank vector is an eigenvector of the transition matrix. ; Where M is the transition matrix and P is the eigenvector of the transition matrix.

[0178] Specifically, the PowerMethod can be used to calculate the PageRank vector. The specific calculation formula is as follows:

[0179] M ; Where i = 0, 1, 2…; Here, R0 is a random vector, which is the initial feature vector shown above.

[0180] Step S712: After obtaining the PageRank vector of the transition matrix, which is the transaction importance of each object in the target group, the target group is classified by the trained classification model to obtain the probability that the target group belongs to the target type group.

[0181] As shown above, the classification model can be a fully connected neural network. Figure 4 This is a schematic diagram of a two-layer fully connected neural network. The input is the PageRank feature vector (the feature vector after sorting). After processing by the intermediate layer, the output is the corresponding class probability vector.

[0182] Step S713: If the probability that the target group belongs to the target type group is greater than the preset probability threshold, that is, the target group belongs to the target type group, which may belong to some abnormal type groups. Then, there may be some abnormal accounts or objects in these abnormal groups. Therefore, after the probability that the target group belongs to the target type group is greater than the preset probability threshold, the objects in the target group whose transaction importance is greater than the preset importance threshold are identified as target objects.

[0183] The preset importance threshold can be either pre-stored in an electronic device by the administrator or entered by the administrator through a terminal device.

[0184] Furthermore, the probability of a target group belonging to a target type group is obtained by performing group classification processing on the trained classification model.

[0185] Based on the same principle as the social network group detection method provided in the embodiments of this application, the embodiments of this application also provide a device for social network group detection, such as... Figure 8 As shown, the social network group detection device 80 may specifically include: an acquisition module 81, a first calculation module 82, a second calculation module 83, a third calculation module 84, and a group classification processing module 85, wherein... The acquisition module 81 is used to acquire the transaction data volume of each transaction between each object in the target group; The first calculation module 82 is used to calculate, for each transaction initiator in each object, a first total transaction data volume between the transaction initiator and each transaction receiver corresponding to the transaction initiator, based on the transaction data volume of each transaction initiated by the transaction initiator, and to calculate a second total transaction data volume of the transactions initiated by the transaction initiator. The second calculation module 83 is used to group a transaction initiator and a transaction receiver in each object into an object group. For each object group, the ratio between the first total transaction data volume corresponding to the object group and the second total transaction volume corresponding to the transaction initiator in the object group is calculated. The third calculation module 84 is used to take the ratio corresponding to each object group as the degree of influence of the transaction initiator on the transaction recipient in that object group, and to calculate the transaction importance of each object in the target group for the ratio corresponding to each object group in the target group. The group classification processing module 85 is used to classify the target group based on the transaction importance of each object in the target group, and obtain the probability that the target group belongs to the target type of group.

[0186] In one possible implementation of this application embodiment, when the third calculation module 94 calculates the transaction importance of each object in the target group based on the ratio corresponding to each object group in the target group, it is specifically used for: Based on the first number of objects in the target group and the ratio corresponding to each object group, a first matrix is ​​constructed in which the number of rows and columns are equal to the first number, and the element value at any position is equal to the ratio corresponding to the object group at that position; wherein, the number of rows and columns of the first matrix are the number of objects contained in the target group, and the element at any position in the first matrix is ​​the ratio corresponding to the object group represented by that position. Construct an initial feature vector, and based on the initial feature vector and the first matrix, calculate the feature vector of the first matrix, where the feature vector of the first matrix is ​​used to characterize the importance of each object in the target group in the transaction of the target group.

[0187] In another possible implementation of this application embodiment, when the group classification processing module 95 performs group classification processing on the target group based on the transaction importance of each object in the target group and obtains the probability that the target group belongs to the target type group, it is specifically used for: The eigenvectors of the first matrix are sorted according to a preset order to obtain the sorted eigenvectors. Based on the sorted feature vectors, and through the trained classification model, group classification is performed to obtain the probability that the target group belongs to the target type group.

[0188] In another possible implementation of this application embodiment, when the third calculation module 94 calculates the transaction importance of each object in the target group based on the ratio corresponding to each object group in the target group, it is specifically used for: Obtain the directed graph of the target transaction reasons corresponding to the target transaction group; where a node in the directed graph of the target transaction reasons represents an object, there is an edge between the nodes corresponding to the two objects that have a transaction, the direction of the edge is from the node of the transaction initiator to the node of the transaction receiver, the weight of the edge is used to represent the reliability of the transaction relationship, and the weight of the edge is directly proportional to the reliability of the transaction relationship. Each object group consists of a transaction initiator and a transaction receiver. For each object group in the target group, the weight of the edges corresponding to that object group is determined from the directed graph of the target transaction reasons. Based on the weight of the edges corresponding to the object group, adjust the ratio between the first total transaction data volume corresponding to the object group and the second total transaction volume corresponding to the transaction initiator in the object group. In this case, the weight of each edge in the directed graph of the transaction reason is inversely proportional to the ratio of the object group corresponding to that edge. Based on the adjusted ratios of each object group, the transaction importance of each object in the target group is calculated.

[0189] Another possible implementation of this application embodiment, the method for obtaining the directed graph of target transaction reasons corresponding to the target transaction group, includes: Obtain the relationship graph between objects in the target group; where the weight of the edge between any two objects represents the degree of intimacy between the two objects, and the weight of the edge between any two objects is directly proportional to the degree of intimacy between the two objects; Obtain chat data between various objects in the target group, and extract the context chat data corresponding to the transactions between each object from the chat data; Based on the contextual chat data corresponding to each object when a transaction occurs, a first transaction reason directed graph is constructed; in the first transaction reason directed graph, a node represents an object, there is an edge between the nodes corresponding to the two objects that have a transaction, the direction of the edge is from the node of the transaction initiator to the node of the transaction receiver, the weight of the edge is used to represent the reliability of the transaction relationship, and the weight of the edge is directly proportional to the reliability of the transaction relationship. Based on the relationship graph between objects in the target group, the weights of each edge in the first directed graph of transaction reasons are adjusted to obtain the second directed graph of transaction reasons. At least one of the first and second directed graphs of transaction reasons is determined as the target transaction reason directed graph corresponding to the target transaction group.

[0190] Another possible implementation of this application embodiment is that the second total transaction data volume corresponding to each transaction initiator in each object is obtained in the following way: Treat each object in the target group as a node, construct an edge between the nodes corresponding to the objects that have a transaction, pointing from the node of the transaction initiator to the node of the transaction receiver, and use the first total transaction data volume between the objects as the weight to construct a directed graph of transaction relationships corresponding to the target group. For each source node in the directed graph of transaction relationships, the second total transaction data volume of the transaction initiator corresponding to that source node is obtained by calculating the sum of the weights of all the edges emanating from that source node.

[0191] In another possible implementation of this application embodiment, the apparatus 90 includes: a first determining module, wherein... The first determining module is used to determine whether the second total transaction volume corresponding to the transaction initiator in the object group is greater than or equal to a preset transaction volume threshold. Furthermore, the device 80 also includes: a fourth computing module, wherein, The fourth calculation module is used to calculate the ratio between the first total transaction volume of the object group and the preset transaction volume threshold when the total transaction volume of the transaction initiator in the object group is less than the preset transaction volume threshold, and to use the ratio as the ratio between the first total transaction data volume of the object group and the second total transaction volume of the transaction initiator in the object group.

[0192] Another possible implementation of this application embodiment is that the transaction data volume of a single transaction includes at least one of the following: transaction resource volume and number of transactions; If the transaction data volume of a single transaction includes: transaction resource volume and transaction count, then when the second calculation module calculates the ratio between the first total transaction data volume corresponding to each object group and the second total transaction volume corresponding to the transaction initiator in that object group, it is specifically used for: Calculate the first ratio between the first total transaction resource amount corresponding to the object group and the second total transaction resource amount corresponding to the transaction initiator in the object group; Calculate the second ratio between the first total number of transactions corresponding to the object group and the second total number of transactions corresponding to the transaction initiators in the object group; Obtain the target weight information corresponding to the first ratio and the second ratio respectively; Based on the first ratio, the second ratio, and their respective weight information, calculate the ratio between the first total transaction data volume corresponding to the object group and the second total transaction volume corresponding to the transaction initiator in the object group.

[0193] In another possible implementation of this application embodiment, for each object group, the second calculation module 93, when obtaining the target weight information corresponding to the first ratio and the second ratio respectively, is specifically used for: Obtain the first weight information corresponding to the first ratio and the second ratio in the object group. The first weight information is obtained by weighting the number of objects contained in the target group, the first ratio and the second ratio through a weight model. Obtain the transaction habit characteristics of the transaction initiators in the object group, as well as the time interval between two adjacent transactions in the object group; Based on trading habits and the time interval between two adjacent transactions, the first weight information is adjusted to obtain the target weight information corresponding to the first ratio and the second ratio.

[0194] In another possible implementation of this application embodiment, if the probability that the target group belongs to the target type group is greater than a preset probability threshold, then the device 80 further includes: a second determining module, wherein, The second determination module is used to determine the objects in the target group whose transaction importance is greater than a preset importance threshold as target objects.

[0195] Another possible implementation of this application embodiment is that the trained classification model is obtained by training the initial model in the following way: Multiple training samples are obtained. Each training sample includes: the amount of transaction data between each object in a sample group, and the label corresponding to the sample group. The label corresponding to a sample group is used to characterize whether the sample group belongs to the target type group. For each sample group, obtain the transaction data volume of each transaction between each object in that sample group; For each transaction initiator in each object, based on the transaction data volume of each transaction initiated by the transaction initiator, calculate the first total transaction data volume between the transaction initiator and each transaction recipient corresponding to the transaction initiator, and calculate the second total transaction data volume of the transactions initiated by the transaction initiator. Each object group consists of a transaction initiator and a transaction receiver. For each object group, the ratio of the first total transaction data volume corresponding to the sample object group to the second total transaction volume corresponding to the transaction initiator in the object group is calculated. The ratio corresponding to each sample object group is used as the degree of influence of the transaction initiator on the transaction recipient in that sample object group. For the ratio corresponding to each object group in that sample group, the transaction importance of each object in that sample group in that sample group is calculated. Based on the importance of each sample object in the transaction within each sample group, and the label corresponding to that sample group, an initial model is trained to obtain a trained classification model.

[0196] It should be noted that the first calculation module 82, the second calculation module 83, the third calculation module 84, and the fourth calculation module can be the same calculation module, different calculation modules, or partially the same calculation module; the first determining module and the second determining module can be the same determining module or different determining modules, and no limitation is made in the embodiments of this application.

[0197] This application provides an apparatus for detecting social network groups. In this embodiment, by acquiring the transaction data volume of each transaction between each object in the target group, and for each transaction initiator in each object, the first total transaction data volume between the transaction initiator and each corresponding transaction recipient of the transaction initiator and the second total transaction data volume of the transaction initiated by the transaction initiator can be calculated. Since a transaction initiator and a transaction recipient in each object are considered as an object group, the ratio between the first total transaction data volume corresponding to each object group and the second total transaction volume corresponding to the transaction initiator in the object group can be obtained based on the first total transaction data volume and the second total transaction data volume. This ratio is used as the degree of influence of the transaction initiator in the object group on the transaction recipient. Based on this ratio, the transaction importance of each object in the target group in the target group can be calculated. Based on the transaction importance of each object in the target group in the target group, the probability of the target group belonging to the target type group can be obtained through a trained classification model, thereby determining whether the target group belongs to the target type group.

[0198] The apparatus in this application embodiment can execute the method provided in this application embodiment, and the implementation principle is similar. The actions performed by each module in the apparatus of each embodiment of this application correspond to the steps in the method of each embodiment of this application. For detailed functional descriptions of each module of the apparatus, please refer to the descriptions in the corresponding methods shown above, which will not be repeated here.

[0199] Figure 9 A schematic diagram of the structure of an electronic device to which this application embodiment applies is shown, such as... Figure 9As shown, the electronic device can be a standalone encoder (such as a general encoder), a standalone decoder (general decoder), or an electronic device that has both encoding and decoding functions. For example, the electronic device can be a server or a user terminal. The electronic device can be used to implement the methods provided in any embodiment of this application.

[0200] like Figure 9 As shown, the electronic device 2000 may primarily include at least one processor 2001. Figure 9 The diagram shows components such as a memory 2002, a communication module 2003, and an input / output interface 2004. Optionally, these components can be connected and communicate with each other via a bus 2005. It should be noted that... Figure 9 The structure of the electronic device 2000 shown is merely illustrative and does not constitute a limitation on the electronic devices to which the methods provided in the embodiments of this application are applicable.

[0201] The memory 2002 can be used to store operating systems and applications, etc. The applications may include computer programs that implement the methods shown in the embodiments of this application when invoked by the processor 2001, and may also include programs for implementing other functions or services. The memory 2002 may be ROM (Read Only Memory) or other types of static storage devices that can store static information and instructions, RAM (Random Access Memory) or other types of dynamic storage devices that can store information and computer programs, or it may be EEPROM (Electrically Erasable Programmable Read Only Memory), CD-ROM (CompactDisc Read Only Memory) or other optical disc storage, optical disk storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium that can be used to carry or store desired program code in the form of instructions or data structures and that can be accessed by a computer, but is not limited thereto.

[0202] Processor 2001 is connected to memory 2002 via bus 2005, and implements corresponding functions by calling application programs stored in memory 2002. Processor 2001 can be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. It can implement or execute the various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of this application. Processor 2001 can also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, etc.

[0203] Electronic device 2000 can connect to a network via communication module 2003 (which may include, but is not limited to, components such as a network interface) to communicate with other devices (such as user terminals or servers) through the network and achieve data interaction, such as sending data to or receiving data from other devices. Communication module 2003 may include wired network interfaces and / or wireless network interfaces, meaning the communication module may include at least one of wired or wireless communication modules.

[0204] Electronic device 2000 can connect to required input / output devices, such as keyboards and display devices, via input / output interface 2004. Electronic device 2000 itself may have a display device, and other display devices can also be connected externally via interface 2004. Optionally, storage devices, such as hard drives, can also be connected via interface 2004 to store data from electronic device 2000, retrieve data from storage devices, or store data from storage devices into memory 2002. It is understood that input / output interface 2004 can be a wired interface or a wireless interface. Depending on the actual application scenario, the device connected to input / output interface 2004 can be a component of electronic device 2000 or an external device connected to electronic device 2000 when needed.

[0205] The bus 2005 used to connect the components may include a pathway for transmitting information between the components. The bus 2005 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus, etc. Depending on its function, the bus 2005 can be divided into address bus, data bus, control bus, etc.

[0206] Optionally, for the solution provided in the embodiments of this application, the memory 2002 can be used to store a computer program that executes the solution of this application, and the processor 2001 runs the computer program. When the processor 2001 runs the computer program, it implements the operation of the method or apparatus provided in the embodiments of this application.

[0207] Based on the same principle as the method provided in the embodiments of this application, the embodiments of this application provide a computer-readable storage medium storing a computer program, which, when executed by a processor, can implement the corresponding content of the aforementioned method embodiments.

[0208] This application also provides a computer program product, which includes a computer program that, when executed by a processor, can implement the corresponding content of the aforementioned method embodiments.

[0209] It should be noted that the terms "first," "second," "third," "fourth," "1," "2," etc. (if present) in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in a sequence other than that shown in the figures or text.

[0210] In the embodiments of this application, the terms "module" or "unit" refer to a computer program or part of a computer program that has a predetermined function and works with other related parts to achieve a predetermined goal, and can be implemented wholly or partially using software, hardware (such as processing circuitry or memory), or a combination thereof. Similarly, a processor (or multiple processors or memory) can be used to implement one or more modules or units. Furthermore, each module or unit can be part of an overall module or unit that includes the functionality of that module or unit.

[0211] It should be understood that although arrows indicate various operation steps in the flowcharts of this application's embodiments, the order in which these steps are implemented is not limited to the order indicated by the arrows. Unless explicitly stated herein, in some implementation scenarios of this application's embodiments, the implementation steps in each flowchart can be executed in other orders as required. Furthermore, some or all steps in each flowchart, based on the actual implementation scenario, may include multiple sub-steps or multiple stages. Some or all of these sub-steps or stages can be executed at the same time, and each sub-step or stage can also be executed at different times. In scenarios where execution times differ, the execution order of these sub-steps or stages can be flexibly configured according to requirements, and this application's embodiments do not limit this.

[0212] The above description is only an optional implementation method for some implementation scenarios of this application. It should be noted that for those skilled in the art, other similar implementation methods based on the technical concept of this application without departing from the technical concept of this application also fall within the protection scope of the embodiments of this application.

Claims

1. A method for detecting social network groups, characterized in that, include: Obtain the transaction data volume of each transaction between each object in the target group; For each transaction initiator among the various objects, based on the transaction data volume of each transaction initiated by the transaction initiator, a first total transaction data volume between the transaction initiator and each transaction recipient corresponding to the transaction initiator is calculated, and a second total transaction data volume of the transactions initiated by the transaction initiator is calculated. Each of the objects is divided into an object group by a transaction initiator and a transaction receiver. For each object group, the ratio between the first total transaction data volume corresponding to the object group and the second total transaction volume corresponding to the transaction initiator in the object group is calculated. The ratio corresponding to each object group is used as the degree of influence of the transaction initiator on the transaction recipient in that object group. For the ratio corresponding to each object group in the target group, the transaction importance of each object in the target group in the target group is calculated. Based on the transaction importance of each object in the target group, the target group is classified using a trained classification model to obtain the probability that the target group belongs to the target type group.

2. The method according to claim 1, characterized in that, The step of calculating the transaction importance of each object in the target group within the target group, based on the ratio corresponding to each object group in the target group, includes: Based on the first number of objects in the target group and the ratio corresponding to each object group, a first matrix is ​​constructed in which the number of rows and columns are equal to the first number, and the element value at any position is equal to the ratio corresponding to the object group at that position; wherein, the number of rows and columns of the first matrix are both the number of objects contained in the target group, and the element at any position in the first matrix is ​​the ratio corresponding to the object group represented by that position. An initial feature vector is constructed, and based on the initial feature vector and the first matrix, the feature vector of the first matrix is ​​calculated, wherein the feature vector of the first matrix is ​​used to characterize the transaction importance of each object in the target group in the target group.

3. The method according to claim 2, characterized in that, The step of classifying the target group based on the transaction importance of each object in the target group using a trained classification model to obtain the probability that the target group belongs to a target type group includes: The eigenvectors of the first matrix are sorted according to a preset order to obtain sorted eigenvectors. Based on the sorted feature vectors, and through the trained classification model, the probability that the target group belongs to the target type group is obtained.

4. The method according to claim 1, characterized in that, The step of calculating the transaction importance of each object in the target group within the target group, based on the ratio corresponding to each object group in the target group, includes: Obtain the target transaction reason directed graph corresponding to the target transaction group; wherein, a node in the target transaction reason directed graph represents an object, there is an edge between the nodes corresponding to the two objects that have a transaction, the direction of the edge is from the node of the transaction initiator to the node of the transaction receiver, the weight of the edge is used to represent the reliability of the transaction relationship, and the weight of the edge is directly proportional to the reliability of the transaction relationship. Take one transaction initiator and one transaction receiver in each of the objects as an object group. For each object group in the target group, determine the weight of the connecting edge corresponding to the object group from the directed graph of the target transaction cause. Based on the weight of the edges corresponding to the object group, the ratio between the first total transaction data volume corresponding to the object group and the second total transaction volume corresponding to the transaction initiator in the object group is adjusted. The weight of each edge in the directed graph of the transaction reason is inversely proportional to the ratio of the object group corresponding to the edge. Based on the adjusted ratios of each object group, the transaction importance of each object in the target group is calculated within the target group.

5. The method according to claim 4, characterized in that, The methods for obtaining the directed graph of the target transaction reasons corresponding to the target transaction group include: Obtain the relationship graph between each object in the target group; wherein, the weight of the edge between any two objects represents the degree of intimacy between the two objects, and the weight of the edge between any two objects is directly proportional to the degree of intimacy between the two objects; Obtain chat data between various objects in the target group, and extract context chat data corresponding to the transaction between each object from the chat data; Based on the contextual chat data corresponding to the transactions of the various objects, a first directed graph of transaction reasons is constructed; wherein, a node in the first directed graph of transaction reasons represents an object, there is an edge between the nodes corresponding to the two objects that have a transaction, the direction of the edge is from the node of the transaction initiator to the node of the transaction receiver, the weight of the edge is used to represent the reliability of the transaction relationship, and the weight of the edge is directly proportional to the reliability of the transaction relationship. Based on the relationship graph between the objects in the target group, the weights of each edge in the first directed graph of transaction reasons are adjusted to obtain the second directed graph of transaction reasons. At least one of the first directed graph of transaction reasons and the second directed graph of transaction reasons is determined as the directed graph of target transaction reasons corresponding to the target transaction group.

6. The method according to claim 1, characterized in that, The second total transaction data volume corresponding to each transaction initiator in each of the aforementioned objects is obtained through the following method: Each object in the target group is treated as a node. An edge is constructed between the nodes corresponding to the objects that have a transaction, pointing from the node of the transaction initiator to the node of the transaction receiver. The first total transaction data volume between the objects is used as the weight to construct a directed graph of the transaction relationship corresponding to the target group. For each source node in the directed graph of the transaction relationship, the second total transaction data volume of the transaction initiator corresponding to the source node is obtained by calculating the sum of the weights of all the edges sent by the source node.

7. The method according to claim 1, characterized in that, Before calculating the ratio between the first total transaction data volume corresponding to the object group and the second total transaction volume corresponding to the transaction initiator in the object group, the method further includes: Determine that the second total transaction volume corresponding to the transaction initiator in the object group is greater than or equal to a preset transaction volume threshold. The method further includes: If the total transaction volume corresponding to the transaction initiator in the object group is less than the preset transaction volume threshold, then the ratio between the first total transaction volume corresponding to the object group and the preset transaction volume threshold is calculated, and this ratio is used as the ratio between the first total transaction data volume corresponding to the object group and the second total transaction volume corresponding to the transaction initiator in the object group.

8. The method according to claim 1, characterized in that, The transaction data volume of a single transaction includes at least one of the following: transaction resource volume and number of transactions; If the transaction data volume of a single transaction includes: transaction resource volume and transaction frequency, then, for each object group, calculating the ratio between the first total transaction data volume corresponding to that object group and the second total transaction volume corresponding to the transaction initiator in that object group includes: Calculate the first ratio between the first total transaction resource amount corresponding to the object group and the second total transaction resource amount corresponding to the transaction initiator in the object group; Calculate the second ratio between the first total number of transactions corresponding to the object group and the second total number of transactions corresponding to the transaction initiators in the object group; Obtain the target weight information corresponding to the first ratio and the second ratio respectively; Based on the first ratio, the second ratio, and their respective weight information, the ratio between the first total transaction data volume corresponding to the object group and the second total transaction volume corresponding to the transaction initiator in the object group is calculated.

9. The method according to claim 8, characterized in that, For each object group, obtaining the target weight information corresponding to the first ratio and the second ratio respectively includes: Obtain the first weight information corresponding to the first ratio and the second ratio in the object group. The first weight information is obtained by weighting the number of objects contained in the target group, the first ratio, and the second ratio through a weight model. Obtain the transaction habit characteristics of the transaction initiators in the object group, as well as the time interval between two adjacent transactions in the object group; Based on the transaction habit characteristics and the time interval between two adjacent transactions, the first weight information is adjusted to obtain the target weight information corresponding to the first ratio and the second ratio respectively.

10. The method according to claim 1, characterized in that, If the probability that the target group belongs to the target type group is greater than a preset probability threshold, the method further includes: Objects in the target group whose transaction importance is greater than a preset importance threshold are identified as target objects.

11. The method according to claim 1, characterized in that, The trained classification model was obtained by training the initial model in the following way: Multiple training samples are obtained, each training sample including: the amount of transaction data between each object in a sample group, and the label corresponding to the sample group. The label corresponding to a sample group is used to characterize whether the sample group belongs to the target type group. For each sample group, obtain the transaction data volume of each transaction between each object in that sample group; For each transaction initiator among the various objects, based on the transaction data volume of each transaction initiated by the transaction initiator, a first total transaction data volume between the transaction initiator and each transaction recipient corresponding to the transaction initiator is calculated, and a second total transaction data volume of the transactions initiated by the transaction initiator is calculated. Each of the objects is divided into an object group by a transaction initiator and a transaction receiver. For each object group, the ratio between the first total transaction data volume corresponding to the sample object group and the second total transaction volume corresponding to the transaction initiator in the object group is calculated. The ratio corresponding to each of the sample object groups is used as the degree of influence of the transaction initiator on the transaction recipient in that sample object group. For the ratio corresponding to each object group in the sample group, the transaction importance of each object in the sample group is calculated in the sample group. The initial model is trained based on the importance of each sample object in the transaction within that sample group and the label corresponding to that sample group, in order to obtain the trained classification model.

12. A device for detecting social network groups, characterized in that, include: The acquisition module is used to acquire the transaction data volume of each transaction between various objects in the target group; The first calculation module is used to calculate, for each transaction initiator among the various objects, a first total transaction data volume between the transaction initiator and each transaction receiver corresponding to the transaction initiator, based on the transaction data volume of each transaction initiated by the transaction initiator, and to calculate a second total transaction data volume of the transactions initiated by the transaction initiator. The second calculation module is used to group one transaction initiator and one transaction receiver in each of the objects into an object group. For each object group, the module calculates the ratio between the first total transaction data volume corresponding to the object group and the second total transaction volume corresponding to the transaction initiator in the object group. The third calculation module is used to take the ratio corresponding to each object group as the degree of influence of the transaction initiator on the transaction recipient in that object group, and to calculate the transaction importance of each object in the target group in the target group for the ratio corresponding to each object group in the target group. The group classification processing module is used to classify the target group based on the transaction importance of each object in the target group using a trained classification model, and to obtain the probability that the target group belongs to the target type of group.

13. An electronic device, characterized in that, The electronic device includes a memory and a processor, the memory storing a computer program, and the processor executing the method for detecting social network groups according to any one of claims 1 to 11 when running the computer program.

14. A computer-readable storage medium, characterized in that, The storage medium stores a computer program that, when executed by a processor, implements the method for detecting social network groups as described in any one of claims 1 to 11.

15. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the method for detecting social network groups as described in any one of claims 1 to 11.

Images