Fake user identification method, system, storage medium and electronic device

By filtering subnets, calculating embedded vectors and clustering processing in social networks, identifying false users, the problem of low accuracy of false user identification in the prior art is solved, and more efficient false user identification is achieved.

CN118364321BActive Publication Date: 2024-11-29PEKING UNIV
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202410781891.2
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-06-18
Publication Date
2024-11-29
Estimated Expiration
2044-06-18

AI Technical Summary

Technical Problem

The prior art has low accuracy when identifying fake users, especially when the fake user behavior is complex and simulates real user behavior, it is difficult to be identified.

Method used

By filtering the subnetwork composed of target posts and their interactive users from the social network graph, the embedding vectors of each user node in the subnet are calculated, and false users are identified using clustering processing.

Benefits of technology

It improves the accuracy of false user identification, can quickly discover false user groups, and improves the overall performance of the identification system.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118364321B_ABST
    Figure CN118364321B_ABST
Patent Text Reader

Abstract

The present application discloses a method, system, storage medium and electronic device for identifying fake users. The method includes: filtering out a subnetwork consisting of a target post and all users interacting with the target post from a social network graph; calculating the embedding vector of each user node in the subnetwork according to the meta-path of the subnetwork; clustering each user node in the subnetwork according to the embedding vector of each user node to obtain a clustering result; when the number of user nodes pre-marked as fake users in the clustering result is greater than a preset threshold, all user nodes in the clustering result are determined as fake users. Therefore, the present application identifies fake users by calculating the embedding vector of each user node in the subnetwork. The embedding vector can characterize the similarity between the user node and the ordinary user node and the fake user node. Therefore, it is possible to quickly discover fake user groups after clustering based on the embedding vector, thereby improving the accuracy of fake user identification.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to a method, system, storage medium and electronic device for identifying fake users. Background Art

[0002] With the rapid development of social network related applications, more and more people are transferring their activities to social networks, but there are currently a large number of fake users in social networks. Fake users usually fuel the spread of online information, attack or top posts to attack the target object or increase the influence of the target object, so it is very important to accurately identify fake users and deal with them.

[0003] In the related art, the identification of fake users mainly involves finding fake users based on fake comments, and then finding fake user groups based on simple flow relationships between fake users. However, as the behavior of fake users becomes more and more complex, and fake users now simulate the behavior of real users, real fake users will not be identified as fake users, thus reducing the accuracy of fake user identification. Summary of the invention

[0004] The embodiments of the present application provide a method, system, storage medium and electronic device for identifying fake users. In order to have a basic understanding of some aspects of the disclosed embodiments, a simple summary is given below. This summary is not a general review, nor is it intended to identify key / important components or describe the scope of protection of these embodiments. Its only purpose is to present some concepts in a simple form as a preface to the detailed description that follows.

[0005] In a first aspect, an embodiment of the present application provides a method for identifying a fake user, the method comprising:

[0006] From the social network graph, filter out the subnetwork consisting of the target post and all users who interact with the target post. The social network graph is composed of users and posts as nodes and the interactive content as edges between nodes.

[0007] According to the meta-path of the sub-network, the embedding vector of each user node in the sub-network is calculated. The meta-path is the relationship flow path between the target post and the user in the sub-network.

[0008] According to the embedding vector of each user node, each user node in the sub-network is clustered to obtain the clustering result;

[0009] When the number of user nodes pre-marked as false users in the clustering result is greater than a preset threshold, all user nodes in the clustering result are determined to be false users.

[0010] Optionally, before calculating the embedding vector of each user in the sub-network according to the meta-path of the sub-network, the following steps are further included:

[0011] Identify whether each interactive content in the sub-network is false content;

[0012] The user nodes corresponding to the interactive contents belonging to false contents in the sub-network are marked as false users.

[0013] Optionally, based on the meta-path of the sub-network, the embedding vector of each user node in the sub-network is calculated, including:

[0014] Performing linear transformation on the attributes of different dimensions included in each user node in the sub-network to map the attributes of different dimensions to the same vector space, and obtaining the basic vector representation of each user node;

[0015] According to the meta-path of the sub-network, the basis vector representation of each user node is aggregated within the path so that the basis vector representation of each user node contains the local network structure information;

[0016] According to the meta-paths of the sub-network, the basis vector representations of each user node are aggregated between paths, so that the basis vector representations of each user node contain global semantic information from different meta-paths;

[0017] The basic vector representation containing local network structure information and global semantic information is determined as the embedding vector of each user node in the sub-network.

[0018] Optionally, based on the meta-path of the sub-network, the basis vector representation of each user node is aggregated within the path, including:

[0019] Determine the neighboring user nodes of each user node according to the meta-path of the sub-network;

[0020] Capture the structural and semantic information between each user node and its neighboring user nodes;

[0021] The structural and semantic information are fused into the basis vector representation of each user node.

[0022] Optionally, based on the meta-path of the sub-network, the basis vector representation of each user node is aggregated between paths, including:

[0023] The attention mechanism is used to fuse information between the meta-paths of the sub-network to capture the global semantic information of each user node in the sub-network;

[0024] The global semantic information is fused into the basis vector representation of each user node.

[0025] Optionally, clustering is performed on each user node in the subnetwork according to the embedding vector of each user node to obtain a clustering result, including:

[0026] Based on the embedding vector of each user node, calculate the similarity distance between each user node;

[0027] The user nodes whose similarity distance is greater than the preset value are aggregated to obtain the clustering result.

[0028] Optionally, before filtering out the target post and the subnetwork consisting of all users interacting with the target post from the social network graph, the following steps are further included:

[0029] Get the reading and comment data of each post in the social network graph in different time periods;

[0030] Construct the fluctuation parameters of the reading volume and comment volume of each post in different time periods;

[0031] Based on the fluctuation parameters, identify the posts that are attacked or affected;

[0032] Target posts that are attacked or influenced.

[0033] In a second aspect, an embodiment of the present application provides a system for identifying fake users, the system comprising:

[0034] A filtering module is used to filter out the sub-network consisting of the target post and all users who interact with the target post from the social network graph. The social network graph is composed of users and posts as nodes and interactive content as edges between nodes.

[0035] A calculation module, used to calculate the embedding vector of each user node in the sub-network according to the meta-path of the sub-network, where the meta-path is the relationship flow path between the target post and the user in the sub-network;

[0036] The clustering module is used to cluster each user node in the sub-network according to the embedding vector of each user node to obtain a clustering result;

[0037] The determination module is used to determine all user nodes in the clustering result as false users when the number of user nodes pre-marked as false users in the clustering result is greater than a preset threshold.

[0038] In a third aspect, an embodiment of the present application provides a computer storage medium, which stores a plurality of instructions, and the instructions are suitable for being loaded by a processor and executing the above-mentioned method steps.

[0039] In a fourth aspect, an embodiment of the present application provides an electronic device, which may include: a processor and a memory; wherein the memory stores a computer program, and the computer program is suitable for being loaded by the processor and executing the above-mentioned method steps.

[0040] The technical solution provided by the embodiments of the present application may have the following beneficial effects:

[0041] In an embodiment of the present application, the present application identifies fake users by calculating the embedding vector of each user node in the subnetwork. The embedding vector can characterize the similarity between the user node and the ordinary user node and the fake user node. Therefore, the fake user group can be quickly discovered after clustering based on the embedding vector, thereby improving the accuracy of fake user identification.

[0042] It should be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0044] Figure 1 It is a flowchart of a method for identifying a fake user provided in an embodiment of the present application;

[0045] Figure 2 This is a schematic diagram of a process for constructing a social network graph provided in an embodiment of the present application;

[0046] Figure 3 It is a relationship diagram between individual posts and user relationships in a social network graph provided by an embodiment of the present application;

[0047] Figure 4 It is a fluctuation diagram of a fluctuation condition of a fluctuation parameter R2 provided in an embodiment of the present application;

[0048] Figure 5 It is a path flow diagram of a meta-path of a sub-network provided in an embodiment of the present application;

[0049] Figure 6 It is a schematic diagram of a sub-network screening process provided by an embodiment of the present application;

[0050] Figure 7 It is a process diagram of a MAGNN algorithm processing process provided by an embodiment of the present application;

[0051] Figure 8 This is a schematic diagram of a clustering result provided in an embodiment of the present application;

[0052] Fig. 9 It is a structural diagram of a false user identification system provided in an embodiment of the present application;

[0053] Fig.10 It is a structural schematic diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0054] The following description and the accompanying drawings sufficiently illustrate specific embodiments of the present application to enable those skilled in the art to practice them.

[0055] It should be clear that the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in the field without creative work are within the scope of protection of the present application.

[0056] When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present application. Instead, they are only examples of systems and methods consistent with some aspects of the present application as detailed in the attached claims.

[0057] In the description of the present application, it should be understood that the terms "first", "second", etc. are used for descriptive purposes only and should not be understood as indicating or implying relative importance. For those of ordinary skill in the art, the specific meanings of the above terms in the present application can be understood according to the specific circumstances. In addition, in the description of the present application, unless otherwise specified, "multiple" refers to two or more. "And / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. The character " / " generally indicates that the objects associated before and after are in an "or" relationship.

[0058] At present, the identification of fake users mainly involves finding fake users based on fake comments, and then finding fake user groups based on simple flow relationships between fake users.

[0059] The inventor of the present application has noticed that as the behavior of fake users becomes more and more complex, and now fake users will simulate the behavior of real users, so that real fake users will not be identified as fake users, thereby reducing the accuracy of fake user identification.

[0060] In order to solve the problem of low accuracy in false user identification, the inventors of the present application have discovered that false users can be identified by calculating the embedding vector of each user node in the sub-network. The embedding vector can characterize the similarity between the user node and ordinary user nodes and false user nodes. Therefore, false user groups can be quickly discovered after clustering based on the embedding vector, thereby improving the accuracy of false user identification.

[0061] The present application provides a method, system, storage medium and electronic device for identifying fake users to solve the problems existing in the above-mentioned related technical problems. Figure 1 -Attached Figure 8 , a method for identifying fake users provided in an embodiment of the present application is introduced in detail. The method can be implemented by a computer program and can be run on a fake user identification system based on the von Neumann system. The computer program can be integrated into an application or run as an independent tool application.

[0062] See also Figure 1 , which is a flow chart of a method for identifying a fake user according to an embodiment of the present application. Figure 1 As shown, the method of the embodiment of the present application may include the following steps:

[0063] S101, filtering out a subnetwork consisting of a target post and all users who interact with the target post from a social network graph, wherein the social network graph is composed of users and posts as nodes and interactive contents as edges between nodes;

[0064] Among them, posts usually refer to user-generated content, including text, pictures, videos or other media types. Posts can be seen, commented and shared by other users, and are the basic components of social network interaction. Target posts are posts that are attacked or affected. Subnetworks are all users who belong to and interact with target posts, such as all users who like or comment on the target post, divided from the social network graph.

[0065] In the embodiment of the present application, a social network graph consisting of users and posts as nodes and interactive content as edges between nodes can be generated by the following steps, including: collecting various social data of users from social networks; preprocessing various social data; performing data calculation on various preprocessed social data to obtain a social network graph consisting of users and posts as nodes and interactive content as edges between nodes. The social network can be, for example, Weibo, Douyin, etc.

[0066] For example Figure 2 As shown, Figure 2 This is a schematic diagram of the construction process of a social network graph provided by the present application, including data access, data processing process and data calculation process; in data access, the data sources of various social data of users, such as posting data, user data and follow-up data, are collected from the social network, and various social data are stored as raw data; then the data processing process is carried out, and the data processing process includes data cleaning, data integration and data format conversion; finally, the data calculation process is carried out, and the data calculation process includes topic extraction, label calculation and relationship construction, so as to obtain a standard social network graph, and store the social network graph.

[0067] It should be noted that data access methods include web crawlers and commercial API docking, while strictly complying with data privacy policies and security regulations.

[0068] In the embodiment of the present application, unstructured social data can be converted into structured graph data through the construction process of the social network graph, providing a data basis for the identification of fake users.

[0069] For example Figure 3 As shown, Figure 3 It is the relationship between posts and users in a social network graph, including ordinary user sets (U), post sets T, comments, and short-term joint comments; ordinary user sets mainly refer to ordinary users on social media platforms; post sets refer to the content of a post from a user. Posts may be posted by ordinary users or by key opinion leaders (KOLs). KOLs may include celebrities, industry experts, well-known bloggers, etc. Comments refer to users’ comments and replies to a post; short-term joint comments refer to the situation in which multiple users comment on the same post within a short period of time, and these users form a short-term joint comment relationship.

[0070] In some embodiments, the method of determining the target post includes the following steps: obtaining the reading volume and comment volume data of each post in the social network graph in different time periods; constructing the fluctuation parameters of the reading volume and comment volume data of each post in different time periods; identifying the attacked or affected posts according to the fluctuation parameters; and taking the attacked or affected posts as the target posts. Among them, the fluctuation parameter R 2 Fluctuations such as Figure 4 shown.

[0071] Specifically, when identifying attacked or affected posts based on fluctuation parameters, the fluctuation parameters can be analyzed based on the learning experience of historical data of a pre-trained neural network model to determine the attacked or affected posts; or a fluctuation threshold can be set, and posts corresponding to fluctuation parameters that exceed the fluctuation threshold can be used as target posts.

[0072] In an embodiment of the present application, after constructing a social network graph and determining a target post, a subnetwork consisting of the target post and nodes of all users who interact with the target post can be screened out from the social network graph.

[0073] The node scale of social networks is very large, with user nodes reaching billions and posts reaching hundreds of billions. It is not feasible to analyze all users and posts for false comments. The users and posts attacked or influenced by false commentators are targeted and are a very limited set. Therefore, by filtering out subnetworks from the social network graph, the analysis efficiency can be improved. Figure 6 As shown, the screened sub-network reduces the connection paths between T2 and user T.

[0074] S102, calculating the embedding vector of each user node in the sub-network according to the meta-path of the sub-network, where the meta-path is the relationship flow path between the target post and the user in the sub-network;

[0075] The relationship flow path is an ordered sequence of node types and edge types defined on the network model, which describes the composite relationship between the node types involved. Figure 5 As shown, the relationship flow path describes the composite relationship between the node types involved, for example:

[0076] UTU (user-post-user) path, indicating a common comment relationship on the same post;

[0077] UTUTU (user-post-user-post-user) path, which means that users find neighboring users by commenting on posts. If you specify posts, you can find users who have commented on these posts together.

[0078] The UUTUU (user-short-term common comment user-post-short-term common comment user-user) path represents the set of users who commented on a post within a short period of time, and fake commenters may be hiding among them.

[0079] Therefore, the meta-path of the sub-network can be regarded as the high-order proximity between two nodes. By introducing the meta-path of the sub-network, the proximity relationship between two adjacent nodes can be characterized.

[0080] In an embodiment of the present application, before calculating the embedding vector of each user in the sub-network according to the meta-path of the sub-network, it also includes: identifying whether each interactive content in the sub-network is false content; marking the user node corresponding to the interactive content in the sub-network that is false content as a false user.

[0081] Specifically, when identifying whether each interactive content in a sub-network is false content, each interactive content in the sub-network can be input into a pre-trained false content identification model, and a judgment can be made based on the empirical parameters of historical false content learned in advance by the model through machine learning technology to determine the interactive content in the sub-network that is false content.

[0082] The embedding vector of each user node in the computing subnetwork may be calculated using a MAGNN algorithm or other neural network methods.

[0083] In an embodiment of the present application, the process of calculating the embedding vector of each user node in the subnetwork according to the meta-path of the subnetwork specifically includes: linearly transforming the attributes of different dimensions included in each user node in the subnetwork to map the attributes of different dimensions to the same vector space to obtain the basic vector representation of each user node; according to the meta-path of the subnetwork, aggregating the basic vector representation of each user node within the path so that the basic vector representation of each user node contains local network structure information; according to the meta-path of the subnetwork, aggregating the basic vector representation of each user node between paths so that the basic vector representation of each user node contains global semantic information from different meta-paths; and determining the basic vector representation containing local network structure information and global semantic information as the embedding vector of each user node in the subnetwork.

[0084] Specifically, according to the meta-path of the sub-network, the specific process of performing intra-path aggregation on the basic vector representation of each user node includes: determining the neighboring user nodes of each user node according to the meta-path of the sub-network; capturing the structural and semantic information between each user node and its neighboring user nodes; and fusing the structural and semantic information into the basic vector representation of each user node.

[0085] Specifically, according to the meta-path of the sub-network, the specific process of inter-path aggregation of the basic vector representation of each user node includes: using the attention mechanism to fuse information between the meta-paths of the sub-network to capture the global semantic information of each user node in the sub-network; fusing the global semantic information into the basic vector representation of each user node.

[0086] For example Figure 7 As shown in Figure 1, the MAGNN algorithm generates node embedding vectors through node content transformation, intra-meta-path aggregation, and inter-meta-path aggregation. Specifically, MAGNN first applies type-specific linear transformations to project heterogeneous node attributes (dimensions that may be unequal for different node types) into the same latent vector space. Next, MAGNN applies intra-meta-path aggregation with an attention mechanism to each meta-path. During this intra-meta-path aggregation, each target node extracts and combines information from meta-path instances that connect the node to its meta-path-based neighbors. In this way, MAGNN captures the structural and semantic information of heterogeneous graphs from the context of neighboring nodes and the meta-paths in between. After intra-meta-path aggregation, MAGNN further uses an attention mechanism for inter-meta-path aggregation to fuse the latent vectors obtained from multiple meta-paths into the final node embedding. By integrating multiple meta-paths, MAGNN can learn comprehensive semantics in heterogeneous graphs.

[0087] S103, clustering each user node in the sub-network according to the embedding vector of each user node to obtain a clustering result;

[0088] In an embodiment of the present application, each user node in the subnetwork is clustered according to the embedding vector of each user node, and the specific process of obtaining the clustering result includes: calculating the similarity distance between each user node based on the embedding vector of each user node; aggregating user nodes whose similarity distance is greater than a preset value to obtain the clustering result.

[0089] In some embodiments, clustering algorithms such as infomap and Kmeans may be used to cluster nodes.

[0090] S104: When the number of user nodes pre-marked as false users in the clustering result is greater than a preset threshold, all user nodes in the clustering result are determined as false users.

[0091] In some embodiments of the present application, when the number of user nodes that pre-mark false users in the clustering results is less than or equal to a preset threshold, all attribute data of users that are not pre-marked in the clustering results can be obtained, and a comprehensive judgment can be made based on all attribute data of users that are not pre-marked to determine the true identity of the users that are not pre-marked.

[0092] Specifically, the comprehensive judgment method can use machine learning to make judgments or push them to expert clients for manual judgment.

[0093] For example Figure 8 As shown, false review groups can be marked through node aggregation.

[0094] In an embodiment of the present application, the present application identifies fake users by calculating the embedding vector of each user node in the subnetwork. The embedding vector can characterize the similarity between the user node and the ordinary user node and the fake user node. Therefore, the fake user group can be quickly discovered after clustering based on the embedding vector, thereby improving the accuracy of fake user identification.

[0095] The following is a system embodiment of the present application, which can be used to execute the method embodiment of the present application. For details not disclosed in the system embodiment of the present application, please refer to the method embodiment of the present application.

[0096] See also Fig. 9 , which shows a schematic diagram of the structure of a false user identification system provided by an exemplary embodiment of the present application. The false user identification system can be implemented as all or part of an electronic device through software, hardware or a combination of both. The system 1 includes a screening module 10, a calculation module 20, a clustering module 30, and a determination module 40.

[0097] A screening module 10 is used to screen out a subnetwork consisting of a target post and all users who interact with the target post from a social network graph, wherein the social network graph is composed of users and posts as nodes and interactive contents as edges between nodes;

[0098] A calculation module 20, for calculating the embedding vector of each user node in the sub-network according to the meta-path of the sub-network, where the meta-path is the relationship flow path between the target post and the user in the sub-network;

[0099] A clustering module 30, used to perform clustering processing on each user node in the sub-network according to the embedding vector of each user node to obtain a clustering result;

[0100] The determination module 40 is configured to determine all user nodes in the clustering result as false users when the number of user nodes pre-marked as false users in the clustering result is greater than a preset threshold.

[0101] Optionally, the system further includes:

[0102] A false content identification module, used to identify whether each interactive content in the sub-network is false content;

[0103] The fake user marking module is used to mark the user nodes corresponding to the interactive contents belonging to the fake contents in the sub-network as fake users.

[0104] Optionally, the system further includes:

[0105] A data acquisition module is used to obtain the reading volume and comment volume data of each post in the social network graph in different time periods;

[0106] Fluctuation parameter construction module, used to construct fluctuation parameters of the reading volume and comment volume data of each post in different time periods;

[0107] A post identification module is used to identify posts that are attacked or affected based on fluctuation parameters;

[0108] A determination module is used to target posts that are attacked or affected.

[0109] It should be noted that, when the false user identification system provided in the above embodiment executes the false user identification method, only the division of the above functional modules is used as an example. In actual application, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device is divided into different functional modules to complete all or part of the functions described above. In addition, the false user identification system provided in the above embodiment and the false user identification method embodiment belong to the same concept, and the embodiment and implementation process thereof are detailed in the method embodiment, which will not be repeated here.

[0110] The serial numbers of the above-mentioned embodiments of the present application are for description only and do not represent the advantages or disadvantages of the embodiments.

[0111] In an embodiment of the present application, the present application identifies fake users by calculating the embedding vector of each user node in the subnetwork. The embedding vector can characterize the similarity between the user node and the ordinary user node and the fake user node. Therefore, the fake user group can be quickly discovered after clustering based on the embedding vector, thereby improving the accuracy of fake user identification.

[0112] The present application also provides a computer-readable medium having program instructions stored thereon, and when the program instructions are executed by a processor, the false user identification method provided by the above-mentioned various method embodiments is implemented.

[0113] The present application also provides a computer program product comprising instructions, which, when executed on a computer, enables the computer to execute the false user identification method of each of the above method embodiments.

[0114] See also Fig.10 , is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. Fig.10 As shown, the electronic device 1000 may include: at least one processor 1001 , at least one network interface 1004 , a user interface 1003 , a memory 1005 , and at least one communication bus 1002 .

[0115] The communication bus 1002 is used to realize the connection and communication between these components.

[0116] The user interface 1003 may include a display screen (Display) and a camera (Camera). Optionally, the user interface 1003 may also include a standard wired interface and a wireless interface.

[0117] The network interface 1004 may optionally include a standard wired interface or a wireless interface (such as a WI-FI interface).

[0118] Among them, the processor 1001 may include one or more processing cores. The processor 1001 uses various interfaces and lines to connect various parts of the entire electronic device 1000, and executes various functions and processes data of the electronic device 1000 by running or executing instructions, programs, code sets or instruction sets stored in the memory 1005, and calling data stored in the memory 1005. Optionally, the processor 1001 can be implemented in at least one hardware form of digital signal processing (Digital Signal Processing, DSP), field programmable gate array (Field-Programmable Gate Array, FPGA), and programmable logic array (Programmable Logic Array, PLA). The processor 1001 can integrate one or a combination of a central processing unit (Central Processing Unit, CPU), a graphics processing unit (Graphics Processing Unit, GPU) and a modem. Among them, the CPU mainly processes the operating system, user interface and application programs; the GPU is responsible for rendering and drawing the content to be displayed on the display screen; the modem is used to process wireless communications. It can be understood that the above-mentioned modem may not be integrated into the processor 1001, and it can be implemented separately through a chip.

[0119] Among them, the memory 1005 may include a random access memory (Random Access Memory, RAM) and may also include a read-only memory (Read-Only Memory). Optionally, the memory 1005 includes a non-transitory computer-readable storage medium. The memory 1005 can be used to store instructions, programs, codes, code sets or instruction sets. The memory 1005 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for at least one function (such as a touch function, a sound playback function, an image playback function, etc.), instructions for implementing the above-mentioned various method embodiments, etc.; the data storage area may store data involved in the above-mentioned various method embodiments, etc. The memory 1005 may also be optionally at least one storage system located away from the aforementioned processor 1001. As Fig.10 As shown, the memory 1005 as a computer storage medium may include an operating system, a network communication module, a user interface module, and a false user identification application.

[0120] exist Fig.10In the electronic device 1000 shown, the user interface 1003 is mainly used to provide an input interface for the user and obtain data input by the user; and the processor 1001 can be used to call the false user identification application stored in the memory 1005, and specifically perform the following operations:

[0121] From the social network graph, filter out the subnetwork consisting of the target post and all users who interact with the target post. The social network graph is composed of users and posts as nodes and the interactive content as edges between nodes.

[0122] According to the meta-path of the sub-network, the embedding vector of each user node in the sub-network is calculated. The meta-path is the relationship flow path between the target post and the user in the sub-network.

[0123] According to the embedding vector of each user node, each user node in the sub-network is clustered to obtain the clustering result;

[0124] When the number of user nodes pre-marked as false users in the clustering result is greater than a preset threshold, all user nodes in the clustering result are determined to be false users.

[0125] In one embodiment, before calculating the embedding vector of each user in the sub-network according to the meta-path of the sub-network, the processor 1001 further performs the following operations:

[0126] Identify whether each interactive content in the sub-network is false content;

[0127] The user nodes corresponding to the interactive contents belonging to false contents in the sub-network are marked as false users.

[0128] In one embodiment, when the processor 1001 calculates the embedding vector of each user node in the sub-network according to the meta-path of the sub-network, the processor 1001 specifically performs the following operations:

[0129] Performing linear transformation on the attributes of different dimensions included in each user node in the sub-network to map the attributes of different dimensions to the same vector space, and obtaining the basic vector representation of each user node;

[0130] According to the meta-path of the sub-network, the basis vector representation of each user node is aggregated within the path so that the basis vector representation of each user node contains the local network structure information;

[0131] According to the meta-paths of the sub-network, the basis vector representations of each user node are aggregated between paths, so that the basis vector representations of each user node contain global semantic information from different meta-paths;

[0132] The basic vector representation containing local network structure information and global semantic information is determined as the embedding vector of each user node in the sub-network.

[0133] In one embodiment, when the processor 1001 performs path aggregation on the basis vector representation of each user node according to the meta-path of the sub-network, the processor 1001 specifically performs the following operations:

[0134] Determine the neighboring user nodes of each user node according to the meta-path of the sub-network;

[0135] Capture the structural and semantic information between each user node and its neighboring user nodes;

[0136] The structural and semantic information are fused into the basis vector representation of each user node.

[0137] In one embodiment, when the processor 1001 performs path aggregation on the basic vector representation of each user node according to the meta-path of the sub-network, the processor 1001 specifically performs the following operations:

[0138] The attention mechanism is used to fuse information between the meta-paths of the sub-network to capture the global semantic information of each user node in the sub-network;

[0139] The global semantic information is fused into the basis vector representation of each user node.

[0140] In one embodiment, when the processor 1001 performs clustering processing on each user node in the subnetwork according to the embedding vector of each user node to obtain the clustering result, the processor 1001 specifically performs the following operations:

[0141] Based on the embedding vector of each user node, calculate the similarity distance between each user node;

[0142] The user nodes whose similarity distance is greater than the preset value are aggregated to obtain the clustering result.

[0143] In one embodiment, before filtering out the target post and the subnetwork consisting of all users interacting with the target post from the social network graph, the processor 1001 further performs the following operations:

[0144] Get the reading and comment data of each post in the social network graph in different time periods;

[0145] Construct the fluctuation parameters of the reading volume and comment volume of each post in different time periods;

[0146] Based on the fluctuation parameters, identify the posts that are attacked or affected;

[0147] Target posts that are attacked or influenced.

[0148] In an embodiment of the present application, the present application identifies fake users by calculating the embedding vector of each user node in the subnetwork. The embedding vector can characterize the similarity between the user node and the ordinary user node and the fake user node. Therefore, the fake user group can be quickly discovered after clustering based on the embedding vector, thereby improving the accuracy of fake user identification.

[0149] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing related hardware through a computer program, and the program for identifying fake users can be stored in a computer-readable storage medium, and when the program is executed, it can include the processes of the above-mentioned embodiments of the methods. The storage medium can be a disk, an optical disk, a read-only storage memory, or a random access memory, etc.

[0150] The above disclosure is only the preferred embodiment of the present application, which certainly cannot be used to limit the scope of rights of the present application. Therefore, equivalent changes made according to the claims of the present application are still within the scope covered by the present application.

Claims

1. A method for identifying fake users, characterized in that: The method comprises: From a social network graph, a subnetwork consisting of a target post and all users who interact with the target post is selected, wherein the social network graph is composed of users and posts as nodes and interactive content as edges between nodes; wherein, Follow these steps to generate a social network graph, including: Collect various social data of users from social networks; Preprocess various social data; Perform data calculation on various pre-processed social data to obtain a social network graph with users and posts as nodes and interactive content as edges between nodes; Calculate the embedding vector of each user node in the sub-network according to the meta-path of the sub-network, wherein the meta-path is the relationship flow path between the target post and the user in the sub-network; According to the embedding vectors of the user nodes, clustering is performed on the user nodes in the subnetwork to obtain a clustering result; wherein the embedding vector represents the similarity between the user node and the ordinary user node and the fake user node, and the fake user simulates the behavior of the real user; When the number of user nodes pre-marked as false users in the clustering result is greater than a preset threshold, all user nodes in the clustering result are determined as false users; wherein, Before filtering out the target post and the sub-network consisting of all users interacting with the target post from the social network graph, the method further includes: Obtaining data on the number of reads and comments of each post in the social network graph in different time periods; Constructing fluctuation parameters of the reading volume and comment volume data of each post in different time periods; identifying the attacked or affected posts according to the fluctuation parameters; The attacked or affected post is used as the target post; wherein, According to the fluctuation parameters, posts that are attacked or affected are identified, including: By analyzing the fluctuation parameters based on the learning experience of the pre-trained neural network model on historical data, the posts that are attacked or affected can be determined; or a fluctuation threshold can be set, and the posts corresponding to the fluctuation parameters that exceed the fluctuation threshold can be used as target posts.

2. The method according to claim 1, characterized in that Before calculating the embedding vector of each user in the sub-network according to the meta-path of the sub-network, the method further includes: Identify whether each interactive content in the sub-network is false content; The user nodes corresponding to the interactive contents belonging to the false contents in the sub-network are marked as false users.

3. The method according to claim 1, characterized in that The step of calculating the embedding vector of each user node in the sub-network according to the meta-path of the sub-network includes: Performing linear transformation on the attributes of different dimensions included in each user node in the subnetwork to map the attributes of different dimensions to the same vector space, thereby obtaining a basic vector representation of each user node; According to the meta-path of the sub-network, the basis vector representation of each user node is aggregated within the path so that the basis vector representation of each user node contains local network structure information; According to the meta-path of the sub-network, performing inter-path aggregation on the basis vector representation of each user node, so that the basis vector representation of each user node contains global semantic information from different meta-paths; A basic vector representation including the local network structure information and the global semantic information is determined as an embedding vector of each user node in the sub-network.

4. The method according to claim 3, characterized in that The step of performing intra-path aggregation on the basic vector representation of each user node according to the meta-path of the sub-network includes: Determining neighboring user nodes of each user node according to the meta-path of the sub-network; Capturing structural and semantic information between each user node and its neighboring user nodes; The structural and semantic information is fused into the basis vector representation of each user node.

5. The method according to claim 3, characterized in that: The performing path aggregation on the basic vector representation of each user node according to the meta-path of the sub-network includes: Using an attention mechanism to fuse information between the meta-paths of the sub-network to capture the global semantic information of each user node in the sub-network; The global semantic information is integrated into the basis vector representation of each user node.

6. The method according to claim 1, characterized in that The clustering of each user node in the sub-network according to the embedding vector of each user node to obtain a clustering result includes: Calculate the similarity distance between the user nodes based on the embedding vector of each user node; The user nodes whose similarity distance is greater than a preset value are aggregated to obtain a clustering result.

7. A system for identifying fake users, characterized in that: The system comprises: A screening module is used to screen out a subnetwork consisting of a target post and all users who interact with the target post from a social network graph, wherein the social network graph is composed of users and posts as nodes and interactive content as edges between nodes; wherein, Follow these steps to generate a social network graph, including: Collect various social data of users from social networks; Preprocess various social data; Perform data calculation on various pre-processed social data to obtain a social network graph with users and posts as nodes and interactive content as edges between nodes; A calculation module, used for calculating the embedding vector of each user node in the sub-network according to the meta-path of the sub-network, wherein the meta-path is the relationship flow path between the target post and the user existing in the sub-network; A clustering module, configured to perform clustering processing on each user node in the subnetwork according to the embedding vector of each user node to obtain a clustering result; wherein the embedding vector represents the similarity between the user node and the ordinary user node and the fake user node, and the fake user will simulate the behavior of the real user; A determination module, configured to determine all user nodes in the clustering result as false users when the number of user nodes pre-marked as false users in the clustering result is greater than a preset threshold; wherein, The system is also specifically used for: Obtaining data on the number of reads and comments of each post in the social network graph in different time periods; Constructing fluctuation parameters of the reading volume and comment volume data of each post in different time periods; identifying the attacked or affected posts according to the fluctuation parameters; The attacked or affected post is used as the target post; wherein, According to the fluctuation parameters, posts that are attacked or affected are identified, including: By analyzing the fluctuation parameters based on the learning experience of the pre-trained neural network model on historical data, the posts that are attacked or affected can be determined; or a fluctuation threshold can be set, and the posts corresponding to the fluctuation parameters that exceed the fluctuation threshold can be used as target posts.

8. A computer storage medium, characterized in that The computer storage medium stores a plurality of instructions, and the instructions are suitable for being loaded by a processor and executing the method according to any one of claims 1 to 6.

9. An electronic device, characterized in that: include: A processor and a memory; wherein the memory stores a computer program, and the computer program is suitable for being loaded by the processor and executing the method according to any one of claims 1 to 6.

Citation Information

Patent Citations

  • Social network abnormal user detection method and device based on heterogeneous graph neural network

    CN112861967A

  • Heterogeneous graph embedding-based abnormal user detection method

    CN118152846A