A method, apparatus and storage medium for detecting an object set

Abstract

The application discloses a detection method and device of an object set and a storage medium, which can be applied to the fields of a map or a vehicle network. The relationship data of a to-be-processed object set is acquired, and a relationship graph network is determined. Then, walking is performed in the nodes of the relationship graph network according to a preset seed object, so that a diffusion object associated with the preset seed object is obtained, and a subgraph network is connected. Then, a graph neural network is called to perform feature extraction on the subgraph network to obtain target graph features, and a target object set in the to-be-processed object set is determined according to the target graph network. Thus, the detection process of the target type object set is realized. Since the seed object is used for diffusion in the node dimension, and the object set is detected in combination with the similarity of the graph network dimension, the accuracy of the object set detection is improved.

Description

Technical Field

[0001] This application relates to the field of computer technology, and in particular to a method, apparatus, and storage medium for detecting a collection of objects. Background Technology

[0002] With the rapid development of internet technology, all kinds of online products have appeared in people's lives, but there may also be non-compliant scenarios and objects in online products, such as black market gangs.

[0003] Generally, the current common detection method for black market gangs is to build a player relationship network and set a weight threshold for the degree between player nodes. Relationships below the threshold are cut off, thereby dividing the network into several smaller networks. Players in the smaller networks are given behavioral characteristics, and rules are used to determine whether they belong to a black market gang.

[0004] However, as criminal gangs constantly change, relying on fixed rules for judgment may result in missed or false detections, affecting the accuracy of object set detection. Summary of the Invention

[0005] In view of this, this application provides a method for detecting object sets, which can effectively improve the accuracy of object set detection.

[0006] The first aspect of this application provides a method for detecting a set of objects, which can be applied to a system or program in a terminal device that includes a function for detecting a set of objects, specifically including:

[0007] Obtain the relational data corresponding to the set of objects to be processed, and determine the relational graph network based on the relational data;

[0008] The preset seed object is used to traverse the nodes of the relational graph network to obtain the diffusion object associated with the preset seed object;

[0009] The nodes corresponding to the diffusion objects in the relational graph network are connected to obtain a subgraph network;

[0010] The graph neural network is invoked to extract features from the subgraph network to obtain the target graph features;

[0011] Based on the target graph features, a target graph network similar to the subgraph network is obtained by filtering in the relation graph network, and a target object set in the set of objects to be processed is determined according to the target graph network. The object types in the target object set are similar to the types of the preset seed objects.

[0012] Optionally, in some possible implementations of this application, obtaining the relational data corresponding to the set of objects to be processed, and determining the relational graph network based on the relational data, includes:

[0013] Obtain the interaction data and position data between each object in the set of objects to be processed;

[0014] The relationship data is obtained by traversing the data based on the correlation between the interaction data and the location data.

[0015] The relational data are linked using triples to obtain a relational chain;

[0016] The relationship chains are integrated to determine the relationship graph network.

[0017] Optionally, in some possible implementations of this application, the step of traversing the nodes of the relational graph network according to a preset seed object to obtain a diffusion object associated with the preset seed object includes:

[0018] The preset seed object is used to roam through the nodes of the relationship graph network to determine various roaming relationships;

[0019] The walk parameters corresponding to each walk relationship are calculated based on the restarted random walk algorithm until the walk parameters converge, and then the walk score vector corresponding to the walk relationship is determined.

[0020] The correlation is calculated based on the wandering score vector to obtain the degree of suspicion;

[0021] If the suspicion level reaches a preset threshold, then the node corresponding to the wandering relationship is determined to be a diffusion object associated with the preset seed object.

[0022] Optionally, in some possible implementations of this application, the method further includes:

[0023] Repeated walks are performed at the nodes of the relational graph network to obtain object update parameters;

[0024] If the object update parameter indicates that the diffusion object has not increased, then the traversal of nodes in the relational graph network is stopped.

[0025] Optionally, in some possible implementations of this application, the step of invoking a graph neural network to extract features from the subgraph network to obtain target graph features includes:

[0026] The preset seed object and the diffusion object are marked as black samples, and the objects in the relationship graph network other than the preset seed object and the diffusion object are marked as white samples;

[0027] A preset neural network configured based on a nested formula of depth maps is invoked, and the black sample and the white sample are input into the preset neural network to obtain classification loss information;

[0028] Based on the classification loss information, the preset neural network is trained using the backpropagation algorithm to obtain the graph neural network;

[0029] The subgraph network is input into the graph neural network for feature extraction to obtain the target graph features.

[0030] Optionally, in some possible implementations of this application, the step of filtering the relational graph network based on the target graph features to obtain a target graph network similar to the subgraph network, and determining the target object set in the set of objects to be processed based on the target graph network, includes:

[0031] Based on the target graph features, the similarity parameters are obtained by traversing the relational graph network.

[0032] The target graph network, which is similar to the subgraph network, is obtained by filtering based on the similarity parameters.

[0033] The target graph network contains objects and the diffusion objects are predicted and weighted to obtain a merged suspicion parameter;

[0034] The target object set in the set of objects to be processed is determined based on the merged suspicion parameter.

[0035] Optionally, in some possible implementations of this application, the detection method for the object set is applied to a blockchain device, where the blockchain device is a node in the blockchain.

[0036] A second aspect of this application provides a detection apparatus for a set of objects, comprising:

[0037] The acquisition unit is used to acquire relational data corresponding to the set of objects to be processed, so as to determine the relational graph network based on the relational data;

[0038] The walking unit is used to walk through the nodes of the relational graph network according to the preset seed object in order to obtain the diffusion object associated with the preset seed object;

[0039] A detection unit is used to connect the nodes corresponding to the diffusion object in the relational graph network to obtain a subgraph network;

[0040] The detection unit is also used to call a graph neural network to extract features from the subgraph network in order to obtain the target graph features;

[0041] The detection unit is further configured to filter out a target graph network similar to the subgraph network in the relation graph network based on the target graph features, and determine a target object set in the set of objects to be processed based on the target graph network, wherein the object type in the target object set is similar to the type of the preset seed object.

[0042] Optionally, in some possible implementations of this application, the acquisition unit is specifically used to acquire the interaction data and position data between each object in the set of objects to be processed;

[0043] The acquisition unit is specifically used to traverse based on the correlation between the interaction data and the location data to obtain the relationship data;

[0044] The acquisition unit is specifically used to associate the relation data in the form of triples to obtain a relation chain;

[0045] The acquisition unit is specifically used to integrate the relationship chains to determine the relationship graph network.

[0046] Optionally, in some possible implementations of this application, the walking unit is specifically used to walk through the nodes of the relation graph network according to the preset seed object in order to determine multiple walking relationships;

[0047] The walking unit is specifically used to calculate the walking parameters corresponding to each walking relationship based on the restarting random walk algorithm, until the walking parameters converge, and then determine the walking score vector corresponding to the walking relationship.

[0048] The walking unit is specifically used to perform correlation calculation based on the walking score vector to obtain the degree of suspicion;

[0049] The walking unit is specifically used to determine that the node corresponding to the walking relationship is a diffusion object associated with the preset seed object if the suspicion level reaches a preset threshold.

[0050] Optionally, in some possible implementations of this application, the walking unit is specifically used to repeatedly walk through the nodes of the relational graph network to obtain object update parameters;

[0051] The walking unit is specifically used to stop walking among the nodes of the relationship graph network if the object update parameter indicates that the diffusion object has not increased.

[0052] Optionally, in some possible implementations of this application, the detection unit is specifically used to mark the preset seed object and the diffusion object as black samples, and to mark the objects in the relationship graph network other than the preset seed object and the diffusion object as white samples;

[0053] The detection unit is specifically used to call a preset neural network configured based on a nested formula of depth map, and input the black sample and the white sample into the preset neural network to obtain classification loss information;

[0054] The detection unit is specifically used to train the preset neural network based on the classification loss information using a backpropagation algorithm to obtain the graph neural network;

[0055] The detection unit is specifically used to input the subgraph network into the graph neural network for feature extraction in order to obtain the target graph features.

[0056] Optionally, in some possible implementations of this application, the detection unit is specifically used to traverse the relational graph network based on the target graph features to obtain similarity parameters;

[0057] The detection unit is specifically used to filter out the target graph network that is similar to the subgraph network based on the similarity parameter;

[0058] The detection unit is specifically used to perform prediction weighting based on the objects contained in the target graph network and the diffusion objects to obtain a merged suspicion parameter;

[0059] The detection unit is specifically used to determine the target object set in the set of objects to be processed based on the merged suspiciousness parameter.

[0060] A third aspect of this application provides a computer device, comprising: a memory, a processor, and a bus system; the memory is used to store program code; the processor is used to execute the object set detection method described in the first aspect or any one of the first aspects according to instructions in the program code.

[0061] A fourth aspect of this application provides a computer-readable storage medium storing instructions that, when executed on a computer, cause the computer to perform the method for detecting a set of objects as described in the first aspect or any one of the first aspects.

[0062] According to one aspect of this application, a computer program product or computer program is provided, comprising computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the object set detection method provided in the first aspect or various optional implementations of the first aspect.

[0063] As can be seen from the above technical solutions, the embodiments of this application have the following advantages:

[0064] By acquiring the relational data corresponding to the set of objects to be processed, a relational graph network is determined based on the relational data. Then, based on a preset seed object, the graph network is navigated through the nodes to obtain diffusion objects associated with the preset seed object. The nodes corresponding to the diffusion objects in the relational graph network are connected to obtain a subgraph network. Then, a graph neural network is invoked to extract features from the subgraph network to obtain target graph features. Subsequently, based on the target graph features, a target graph network similar to the subgraph network is obtained through filtering in the relational graph network. Based on the target graph network, a target object set in the set of objects to be processed is determined. The object types in this target object set are similar to the preset seed object types. This achieves the detection process of the target type object set. By using seed objects for node-level diffusion and combining graph network-level similarity for object set detection, the accuracy of object set detection is improved by performing set detection from different dimensions. Attached Figure Description

[0065] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of this application. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0066] Figure 1 A network architecture diagram for a detection system running on a collection of objects;

[0067] Figure 2 A flowchart illustrating the detection process of an object set, provided as an embodiment of this application.

[0068] Figure 3 A flowchart illustrating a method for detecting an object set as provided in an embodiment of this application;

[0069] Figure 4 A schematic diagram illustrating a scenario for a method for detecting an object set provided in an embodiment of this application;

[0070] Figure 5 A schematic diagram illustrating a scenario for another object set detection method provided in an embodiment of this application;

[0071] Figure 6 A schematic diagram illustrating a scenario for another object set detection method provided in an embodiment of this application;

[0072] Figure 7 A schematic diagram illustrating a scenario for another object set detection method provided in an embodiment of this application;

[0073] Figure 8 A schematic diagram of the structure of an object set detection device provided in an embodiment of this application;

[0074] Figure 9 This is a schematic diagram of the structure of a terminal device provided in an embodiment of this application;

[0075] Figure 10 This application provides a schematic diagram of the structure of a server according to an embodiment of the present application.

[0076] Figure 11A A data sharing system provided in this application embodiment;

[0077] Figure 11B This application provides an embodiment of a blockchain block composition;

[0078] Figure 11C This application provides input information for a blockchain node in an embodiment. Detailed Implementation

[0079] This application provides a method and related apparatus for detecting object sets, which can be applied to systems or programs in terminal devices that include object set detection functionality. The method involves acquiring relational data corresponding to the object set to be processed, and then determining a relational graph network based on the relational data. Next, a preset seed object is used to navigate through the nodes of the relational graph network to obtain diffusion objects associated with the preset seed object. The nodes corresponding to the diffusion objects in the relational graph network are then connected to obtain a subgraph network. A graph neural network is then invoked to extract features from the subgraph network to obtain target graph features. Furthermore, based on the target graph features, a target graph network similar to the subgraph network is obtained through filtering within the relational graph network. Finally, the target object set in the object set to be processed is determined based on the target graph network, where the object types in the target object set are similar to the preset seed object types. This achieves the detection process for target type object sets. By using seed objects for node-level diffusion and combining graph network-level similarity for object set detection, the accuracy of object set detection is improved by performing set detection from different dimensions.

[0080] The terms “first,” “second,” “third,” “fourth,” 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 particular 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, for example, in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “corresponding to,” and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0081] It should be understood that the object set detection method provided in this application can be applied to systems or programs in terminal devices that include object set detection functionality, such as anomaly detection. Specifically, the object set detection system can run in environments such as... Figure 1 In the network architecture shown, such as Figure 1 The diagram shows the network architecture of the object collection detection system. As can be seen, the system can detect object collections from multiple information sources. This is achieved through interaction with the terminal, triggering the server to perform anomaly detection on the involved objects. It can be understood that... Figure 1 The diagram illustrates various terminal devices, which can be computer devices. In real-world scenarios, more or fewer types of terminal devices may participate in the detection of the object set. The specific number and types depend on the actual scenario and are not limited here. Additionally, Figure 1 The example shows one server, but in real-world scenarios, multiple servers can be involved, especially in scenarios involving multi-model training and interaction. The specific number of servers depends on the actual scenario.

[0082] In this embodiment, the server can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms. The terminal can be a smartphone, tablet, laptop, desktop computer, smart speaker, smartwatch, etc., but is not limited to these. The terminal and server can be directly or indirectly connected via wired or wireless communication, and the terminal and server can be connected to form a blockchain network; this application does not impose any restrictions.

[0083] It is understood that the aforementioned object set detection system can run on personal mobile terminals, such as as an anomaly detection application, or it can run on servers, or it can run on third-party devices to provide object set detection and obtain the detection and processing results of the object set of the information source. Specifically, the object set detection system can run as a program on the aforementioned devices, or it can run as a system component of the aforementioned devices, or it can run as a cloud service program. The specific operating mode depends on the actual scenario and is not limited here.

[0084] With the rapid development of internet technology, all kinds of online products have appeared in people's lives, but there may also be non-compliant scenarios and objects in online products, such as black market gangs.

[0085] Generally, the current common detection method for black market gangs is to build a player relationship network and set a weight threshold for the degree between player nodes. Relationships below the threshold are cut off, thereby dividing the network into several smaller networks. Players in the smaller networks are given behavioral characteristics, and rules are used to determine whether they belong to a black market gang.

[0086] However, as criminal gangs constantly change, relying on fixed rules for judgment may result in missed or false detections, affecting the accuracy of object set detection.

[0087] To address the aforementioned problems, this application proposes a method for detecting object sets, which is applied to... Figure 2 In the workflow framework for detecting the object set shown, such as Figure 2 The diagram shown illustrates a process architecture for detecting object sets according to an embodiment of this application. Through interactive operations on the terminal, the server collects relevant interactive information, constructs relationship chains, and segments the relationship chains indicating target types, thereby obtaining the corresponding object sets. This process integrates a diffusion network and a deep graph nesting algorithm for detecting black market groups. The solution integrates a restarted random walk algorithm and a deep graph nesting algorithm, performing deep modeling of the object relationship network and detecting object sets based on network features.

[0088] It is understood that the method provided in this application can be a program written as processing logic in a hardware system, or it can be a device for detecting object sets, implemented in an integrated or external manner. As one implementation, the object set detection device acquires relational data corresponding to the object set to be processed, and determines a relational graph network based on the relational data; then, it traverses the nodes of the relational graph network according to a preset seed object to obtain diffusion objects associated with the preset seed object; it connects the nodes corresponding to the diffusion objects in the relational graph network to obtain a subgraph network; then, it calls a graph neural network to extract features from the subgraph network to obtain target graph features; furthermore, it filters the relational graph network based on the target graph features to obtain a target graph network similar to the subgraph network, and determines the target object set in the object set to be processed based on the target graph network, wherein the object types in the target object set are similar to the preset seed object types. This achieves the detection process of the target type object set. By using seed objects for node-level diffusion and combining graph network-level similarity for object set detection, it improves the accuracy of object set detection by performing set detection from different dimensions.

[0089] The solutions provided in this application relate to machine learning technology in artificial intelligence, and are specifically illustrated through the following embodiments:

[0090] Based on the above process architecture, the method for detecting object collections in this application will be described below. Please refer to [link / reference]. Figure 3 , Figure 3 The flowchart illustrates a method for detecting an object set provided in this application embodiment. This management method can be executed by a server or a terminal, and this application embodiment includes at least the following steps:

[0091] 301. Obtain the relational data corresponding to the set of objects to be processed, and determine the relational graph network based on the relational data.

[0092] In this embodiment, the object to be processed can be a user, player, terminal, or an indicative identifier. For example, when the object to be processed is a player, player relationship-related data (such as game relationship, team relationship, battle record relationship, friend relationship, etc.) is integrated, and these relationship data are constructed into a relationship graph network.

[0093] Specifically, the acquisition of relational data involves first obtaining the interaction data and location data between each object in the set of objects to be processed. For example, interaction data includes collecting player relationship data such as team relationships, match relationships, friend relationships, communication relationships, kill relationships, and support relationships, while location data includes geographical relationships and server relationships. Then, based on the correlation between the interaction data and location data, a traversal is performed to obtain the relational data, which is stored in multiple two-dimensional arrays. For example, team relationships are identified by the following formula, where N is the total number of players globally, and R is the number of players in row a and column b. a，b This represents the number of teams formed between player a and player b.

[0094]

[0095] Furthermore, the relational data is linked in the form of triples to obtain the relational chain. Since there are cross-object relationships between objects, for example, object A and object B are friends, and object B and object C are friends, the relevant data of object B can be written into the relational data between object A and object C. This relational data is processed in the form of quintuples. Therefore, the relational data is processed in the form of N-tuples, which is not limited here.

[0096] Specifically, this involves merging team relationships, match relationships, and friend relationships into an N-tuple-based relationship chain. This means integrating relationship data containing the same object across team relationships, match relationships, and friend relationships. The specific formula can be expressed as:

[0097] RL = (R team ,R battle …R friend )

[0098] Understandably, since each relation is a 2D matrix of size N×N, the relation chains are integrated to determine the relation graph network; that is, the relation chains can also be represented as:

[0099]

[0100] Where K is the number of relation types, RL a,b,c This represents the weights of player b and player c in relation a.

[0101] In one possible scenario, a relational graph network is like... Figure 4 As shown, Figure 4This is a schematic diagram illustrating a method for detecting an object set according to an embodiment of this application. The diagram shows the node distribution of a relational graph network, from which the target object set is extracted in this embodiment. The specific target object set includes black market gangs and individuals involved in black market activities. Examples include black market gangs involved in score boosting, score deduction, and game boosting, and also extend to other game genres such as smoba games and FPS games. The specific type of object set and the type of application depend on the actual scenario. This example uses black market gangs for illustration and is not intended to limit the scope.

[0102] 302. Based on the preset seed object, walk through the nodes of the relational graph network to obtain the diffusion object associated with the preset seed object.

[0103] In this embodiment, the Random Walk with Restart (RWR) algorithm can be used to walk around the nodes of the relationship graph network. That is, starting from a small number of known gang players, it spreads to the surrounding areas of the relationship graph network and predicts new malicious gang players.

[0104] Specifically, the process of walking among nodes in a relational graph network involves first identifying various walking relationships based on a pre-defined seed object (e.g., a user marked as a black market user). Then, a restarted random walk algorithm is used to calculate the walking parameters for each relationship until convergence. Specifically, the restarted random walk algorithm starts from a node in the relational graph network, with each step presenting two choices: randomly select an adjacent node or return to the starting node. During the walk, the restart probability parameter *c* is iteratively calculated. Here, 1-c represents the probability of moving to an adjacent node, and the walking parameters are the weighted sum of *c* and 1-c. The process reaches stationarity after iteration, indicating convergence of the walking parameters. The resulting probability distribution can be considered as a distribution influenced by the starting node. This restarted random walk can capture various relationships between two nodes, revealing the overall structural information of the relational graph network.

[0105] Furthermore, after the walking parameters converge, the walking score vector corresponding to the walking relation can be determined; whereby the construction of the walking score vector r′ can be expressed as:

[0106]

[0107] in,

[0108]

[0109] Using the restarted random walk algorithm (as shown below), r′ is continuously updated until r′ converges:

[0110] r′=cWr′+(1-c)

[0111] Where c is the restart probability, W is the standardized relation matrix, i.e., W = normalized(R), and E is the identity matrix.

[0112] Specifically, the convergence result is as follows:

[0113] r′=(1-c)(E-cW) -1

[0114] The converged r′ walk score vector is the desired score vector r.

[0115] Furthermore, correlation calculations are performed based on the walk score vector to obtain the suspicion level. If the suspicion level reaches a preset threshold, the node corresponding to the walk relationship is determined to be a diffusion object associated with the preset seed object. That is, the suspicion level s of each player being a member of a criminal gang can be calculated using the following formula. i×j :

[0116]

[0117] Where i represents the player object whose suspiciousness is being calculated, and j is a known black market object. By calculating the similarity between i and j in N dimensions, we can obtain the suspiciousness of R players as members of a black market gang.

[0118] After obtaining the suspicion level of each player, by comparing it with the suspicion level s i By comparing the threshold, it can be determined whether player i is a member of a criminal gang, and by linking all the suspicious criminal gangs in the relationship chain, the criminal gang can be found.

[0119] Alternatively, the above walking steps can be repeated until the number of newly predicted gang players no longer increases, that is, by repeatedly walking through the nodes of the relationship graph network to obtain object update parameters; if the object update parameters indicate that the number of diffused objects has not increased, then the walking through the nodes of the relationship graph network is stopped, thereby improving the integrity of the diffused objects.

[0120] 303. Connect the nodes corresponding to the diffusion objects in the relational graph network to obtain the subgraph network.

[0121] In this embodiment, a subgraph network is constructed by connecting the nodes where the diffusion objects (i.e., those indicating potential malicious players) are located in the relationship graph network. Considering that malicious players have the characteristics of gangs, that is, multiple malicious players are associated in the form of black market gangs, the construction of the subgraph network can perform black market gang-level object detection from the graph network dimension. This can filter the associations between multiple malicious players, improve the accuracy of black market gang detection, and better suit the scenario of multi-player detection, thereby enabling the subsequent object detection process from the graph network dimension.

[0122] 304. Call the graph neural network to extract features from the subgraph network to obtain the target graph features.

[0123] In this embodiment, the graph neural network (GCN) can be trained based on a preset seed object and a diffusion object.

[0124] For the training process, firstly, the preset seed objects and diffusion objects are marked as black samples, and the objects in the relational graph network other than the preset seed objects and diffusion objects are marked as white samples. That is, the training labels of player nodes identified as black groups are recorded as black, and the transitions of other player nodes are recorded as white. Then, the preset neural network configured based on the nested formula of the depth graph is called; and the black samples and white samples are input into the preset neural network to obtain classification loss information; then, based on the classification loss information, the backpropagation algorithm is used to train the preset neural network to obtain the graph neural network.

[0125] Specifically, the process of constructing a graph neural network using the Deep Graph Concatenation (GCN) algorithm is shown below:

[0126]

[0127] in, Represents the characteristic of node i in layer l; σ represents the nonlinear transformation; C ij N represents the normalization factor; i W represents all neighbors of node i, including node i itself; l) b represents the weight of the l-th layer; (l) This represents the intercept of the l-th layer.

[0128] Then, let Z be the output of the last layer of GCN, and L be the classification loss of GCN, which can be expressed as:

[0129]

[0130] Among them, Y L F represents the set of nodes with class labels, where F represents the number of classes.

[0131] Then, the backpropagation algorithm of the neural network is used to train the GCN, and the black market labels of all players on the network are predicted again. That is, the subgraph network is input into the graph neural network to extract features to obtain the target graph features.

[0132] 305. Based on the features of the target graph, filter the relational graph network to obtain a target graph network similar to the subgraph network, and determine the target object set in the set of objects to be processed based on the target graph network.

[0133] Then, let Z be the output of the last layer of GCN, and L be the classification loss of GCN, which can be expressed as:

[0134]

[0135] Among them, Y L F represents the set of nodes with class labels, where F represents the number of classes.

[0136] Then, the backpropagation algorithm of the neural network is used to train the GCN, and the black market labels of all players on the network are predicted again. That is, the subgraph network is input into the graph neural network to extract features to obtain the target graph features.

[0137] 305. Based on the features of the target graph, filter the relational graph network to obtain a target graph network similar to the subgraph network, and determine the target object set in the set of objects to be processed based on the target graph network.

[0138] In this embodiment, the object types in the target object set are similar to the types of the preset seed objects, that is, the black market group can be obtained, and the black market players or objects in it can be obtained.

[0139] Furthermore, the process of determining the set of target objects in the set of objects to be processed based on the target graph network can be obtained by weighted merging of prediction results. That is, firstly, the relationship graph network is traversed based on the features of the target graph to obtain similarity parameters; then, the target graph network similar to the subgraph network is obtained by filtering based on the similarity parameters; and finally, the predictions are weighted based on the objects contained in the target graph network and the diffused objects to obtain the merging suspicion parameter. In other words, the prediction results are weighted and merged to obtain the merging player suspicion level.

[0140] s total =a*s RWR +b*s GCN

[0141] Where a and b are the weights of the RWR prediction results and the GCN prediction results, respectively.

[0142] Then, based on the merged suspicion level parameter, the target object set in the set of objects to be processed is determined, i.e., the merged suspicion level s. totalPlayers whose numbers exceed a certain threshold are considered to be involved in illegal activities, thus improving the accuracy of detecting such activities.

[0143] Understandably, fixed rules are easily perceived by criminal gangs, and it is difficult to separate low-profile criminal gang networks. In addition, criminal gang networks are usually impure, containing some normal players. Conventional methods are difficult to accurately distinguish between normal players and criminal players. By comparing feature similarity at the node level and graph network level, the influence of the activity level of criminal gangs can be avoided, thus improving the accuracy of detection.

[0144] The following explanation uses the detection of black market groups in game scenarios as an example. Figure 5 As shown, Figure 5 This is a schematic diagram illustrating another method for detecting object sets provided in this application embodiment. First, player relationship-related data (such as match relationships, team relationships, battle record relationships, friend relationships, etc.) is integrated and constructed into a relationship graph network. Then, using the RWR algorithm (Rebooting Random Walk algorithm), starting with a small number of known gang players, the algorithm expands outwards from the relationship graph network, predicting new malicious gang players. This process is repeated until the number of newly predicted gang players no longer increases. Next, nodes containing malicious players are connected in the relationship graph network to construct subgraph networks. The features of the subgraph networks containing malicious players are then constructed using the Deep Graph Neural Network (GCN) algorithm. Finally, subgraph networks with identical features are detected in the entire relationship graph network, and filtering allows for the location of malicious gangs.

[0145] Furthermore, when using the detection method provided in this embodiment, a corresponding interface can also be displayed, such as... Figure 6 As shown, Figure 6 This is a schematic diagram of a scenario for another object set detection method provided in this application embodiment; the diagram shows that the detection method provided in this embodiment can be used to detect black market players or groups in the preparation stage of a shooting game, and display the corresponding abnormal detection information A1, thereby improving the player's gaming experience.

[0146] In one scenario, this embodiment was launched and ran stably in a key FPS game, with core metrics improved by about 90% and anti-cheat related issues returned to stability.

[0147] In addition, it is also possible to perform, such as Figure 6 The display shown Figure 7This is a schematic diagram of a scenario for another object set detection method provided in this application embodiment; the diagram shows that the detection method provided in this embodiment can prompt the player in the current game in a battle game. If there are cheaters, black market players or groups detected by the detection method provided in this embodiment, the interface prompt B1 can be displayed, thereby improving the player's game experience and allowing the user to upload relevant battle materials for relevant personnel to check.

[0148] It is understood that this embodiment can be extended to other types of games, such as smoba games, and is not limited here.

[0149] As described in the above embodiments, by acquiring the relational data corresponding to the set of objects to be processed, a relational graph network is determined based on the relational data. Then, based on a preset seed object, the object traverses the nodes of the relational graph network to obtain diffusion objects associated with the preset seed object. The nodes corresponding to the diffusion objects in the relational graph network are then connected to obtain a subgraph network. A graph neural network is then invoked to extract features from the subgraph network to obtain target graph features. Furthermore, based on the target graph features, a target graph network similar to the subgraph network is obtained through filtering within the relational graph network. The target graph network is then used to determine the target object set within the set of objects to be processed, where the object types are similar to the preset seed object types. This achieves the detection process for target type object sets. By using seed objects for node-level diffusion and combining graph network-level similarity for object set detection, the accuracy of object set detection is improved by performing set detection from different dimensions.

[0150] To better implement the above-described solutions of the embodiments of this application, related apparatus for implementing the above solutions is also provided below. Please refer to... Figure 8 , Figure 8 This is a schematic diagram of the structure of an object set detection device provided in an embodiment of this application. The object set detection device 800 includes:

[0151] The acquisition unit 801 is used to acquire the relational data corresponding to the set of objects to be processed, so as to determine the relational graph network based on the relational data;

[0152] The walking unit 802 is used to walk through the nodes of the relational graph network according to the preset seed object in order to obtain the diffusion object associated with the preset seed object;

[0153] The detection unit 803 is used to connect the nodes corresponding to the diffusion object in the relational graph network to obtain a subgraph network;

[0154] The detection unit 803 is also used to call a graph neural network to extract features from the subgraph network in order to obtain the target graph features;

[0155] The detection unit 803 is further configured to filter out a target graph network similar to the subgraph network in the relation graph network based on the target graph features, and determine a set of target objects in the set of objects to be processed based on the target graph network, wherein the object types in the set of target objects are similar to the types of the preset seed objects.

[0156] Optionally, in some possible implementations of this application, the acquisition unit 801 is specifically used to acquire the interaction data and position data between each object in the set of objects to be processed;

[0157] The acquisition unit 801 is specifically used to traverse based on the correlation between the interaction data and the location data to obtain the relationship data;

[0158] The acquisition unit 801 is specifically used to associate the relation data in the form of triples to obtain a relation chain;

[0159] The acquisition unit 801 is specifically used to integrate the relationship chain to determine the relationship graph network.

[0160] Optionally, in some possible implementations of this application, the walking unit 802 is specifically used to walk through the nodes of the relation graph network according to the preset seed object in order to determine multiple walking relationships;

[0161] The walking unit 802 is specifically used to calculate the walking parameters corresponding to each walking relationship based on the restarted random walk algorithm, until the walking parameters converge, and then determine the walking score vector corresponding to the walking relationship.

[0162] The walking unit 802 is specifically used to perform correlation calculation based on the walking score vector to obtain the degree of suspicion;

[0163] The walking unit 802 is specifically used to determine that the node corresponding to the walking relationship is a diffusion object associated with the preset seed object if the suspicion level reaches a preset threshold.

[0164] Optionally, in some possible implementations of this application, the walking unit 802 is specifically used to repeatedly walk through the nodes of the relational graph network to obtain object update parameters;

[0165] The walking unit 802 is specifically used to stop walking in the nodes of the relationship graph network if the object update parameter indicates that the diffusion object has not increased.

[0166] Optionally, in some possible implementations of this application, the detection unit 803 is specifically used to mark the preset seed object and the diffusion object as black samples, and to mark the objects in the relationship graph network other than the preset seed object and the diffusion object as white samples;

[0167] The detection unit 803 is specifically used to call a preset neural network configured based on a depth map nested formula, and input the black sample and the white sample into the preset neural network to obtain classification loss information;

[0168] The detection unit 803 is specifically used to train the preset neural network based on the classification loss information using a backpropagation algorithm to obtain the graph neural network;

[0169] The detection unit 803 is specifically used to input the subgraph network into the graph neural network for feature extraction in order to obtain the target graph features.

[0170] Optionally, in some possible implementations of this application, the detection unit 803 is specifically used to traverse the relational graph network based on the target graph features to obtain similarity parameters;

[0171] The detection unit 803 is specifically used to filter out the target graph network that is similar to the subgraph network according to the similarity parameter.

[0172] The detection unit 803 is specifically used to perform prediction weighting based on the objects contained in the target graph network and the diffusion objects to obtain the merged suspicion parameter;

[0173] The detection unit 803 is specifically used to determine the target object set in the set of objects to be processed based on the merged suspiciousness parameter.

[0174] By acquiring the relational data corresponding to the set of objects to be processed, a relational graph network is determined based on the relational data. Then, based on a preset seed object, the graph network is navigated through the nodes to obtain diffusion objects associated with the preset seed object. The nodes corresponding to the diffusion objects in the relational graph network are connected to obtain a subgraph network. Then, a graph neural network is invoked to extract features from the subgraph network to obtain target graph features. Subsequently, based on the target graph features, a target graph network similar to the subgraph network is obtained through filtering in the relational graph network. Based on the target graph network, a target object set in the set of objects to be processed is determined. The object types in this target object set are similar to the preset seed object types. This achieves the detection process of the target type object set. By using seed objects for node-level diffusion and combining graph network-level similarity for object set detection, the accuracy of object set detection is improved by performing set detection from different dimensions.

[0175] This application also provides a terminal device, such as... Figure 9 The diagram shown is a structural schematic of another terminal device provided in an embodiment of this application. For ease of explanation, only the parts related to the embodiment of this application are shown. For specific technical details not disclosed, please refer to the method section of the embodiment of this application. The terminal can be any terminal device including mobile phones, tablet computers, personal digital assistants (PDAs), point-of-sale (POS) terminals, in-vehicle computers, etc. Taking a mobile phone as an example:

[0176] Figure 9 This is a block diagram illustrating a portion of the structure of a mobile phone related to the terminal provided in the embodiments of this application. (Reference) Figure 9 The mobile phone includes components such as a radio frequency (RF) circuit 910, a memory 920, an input unit 930, a display unit 940, a sensor 950, an audio circuit 960, a wireless fidelity (WiFi) module 970, a processor 980, and a power supply 990. Those skilled in the art will understand that... Figure 9 The mobile phone structure shown does not constitute a limitation on the mobile phone and may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0177] The following is combined Figure 9 A detailed introduction to each component of a mobile phone:

[0178] RF circuit 910 can be used for receiving and transmitting signals during information transmission or calls. Specifically, it receives downlink information from the base station and processes it with processor 980; additionally, it transmits uplink data to the base station. Typically, RF circuit 910 includes, but is not limited to, an antenna, at least one amplifier, a transceiver, a coupler, a low-noise amplifier (LNA), and a duplexer. Furthermore, RF circuit 910 can also communicate wirelessly with networks and other devices. The aforementioned wireless communication can use any communication standard or protocol, including but not limited to Global System for Mobile Communication (GSM), General Packet Radio Service (GPRS), Code Division Multiple Access (CDMA), Wideband Code Division Multiple Access (WCDMA), Long Term Evolution (LTE), email, and Short Message Service (SMS).

[0179] The memory 920 can be used to store software programs and modules. The processor 980 executes various functions and data processing of the mobile phone by running the software programs and modules stored in the memory 920. The memory 920 may mainly include a program storage area and a data storage area. The program storage area may store the operating system, at least one application program required for a function (such as sound playback function, image playback function, etc.), etc.; the data storage area may store data created according to the use of the mobile phone (such as audio data, phonebook, etc.). In addition, the memory 920 may include high-speed random access memory, and may also include non-volatile memory, such as at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0180] The input unit 930 can be used to receive input numerical or character information, and to generate key signal inputs related to user settings and function control of the mobile phone. Specifically, the input unit 930 may include a touch panel 931 and other input devices 932. The touch panel 931, also known as a touch screen, can collect touch operations performed by the user on or near it (such as operations performed by the user using a finger, stylus, or any suitable object or accessory on or near the touch panel 931, and air touch operations within a certain range on the touch panel 931), and drive the corresponding connection devices according to a pre-set program. Optionally, the touch panel 931 may include two parts: a touch detection device and a touch controller. The touch detection device detects the user's touch position and the signal generated by the touch operation, and transmits the signal to the touch controller; the touch controller receives touch information from the touch detection device, converts it into touch point coordinates, sends it to the processor 980, and can receive and execute commands sent by the processor 980. In addition, the touch panel 931 can be implemented using various types such as resistive, capacitive, infrared, and surface acoustic wave. In addition to the touch panel 931, the input unit 930 may also include other input devices 932. Specifically, other input devices 932 may include, but are not limited to, one or more of the following: physical keyboard, function keys (such as volume control buttons, power buttons, etc.), trackball, mouse, joystick, etc.

[0181] The display unit 940 can be used to display information input by the user or information provided to the user, as well as various menus of the mobile phone. The display unit 940 may include a display panel 941, which may optionally be configured as a liquid crystal display (LCD), organic light-emitting diode (OLED), or similar form. Further, a touch panel 931 may cover the display panel 941. When the touch panel 931 detects a touch operation on or near it, it transmits the information to the processor 980 to determine the type of touch event. Subsequently, the processor 980 provides corresponding visual output on the display panel 941 based on the type of touch event. Although in Figure 9 In this embodiment, the touch panel 931 and the display panel 941 are two separate components to realize the input and output functions of the mobile phone. However, in some embodiments, the touch panel 931 and the display panel 941 can be integrated to realize the input and output functions of the mobile phone.

[0182] The mobile phone may also include at least one sensor 950, such as a light sensor, a motion sensor, and other sensors. Specifically, the light sensor may include an ambient light sensor and a proximity sensor. The ambient light sensor can adjust the brightness of the display panel 941 according to the ambient light level, and the proximity sensor can turn off the display panel 941 and / or backlight when the phone is moved to the ear. As a type of motion sensor, an accelerometer sensor can detect the magnitude of acceleration in various directions (generally three axes). When stationary, it can detect the magnitude and direction of gravity, which can be used for applications that recognize the phone's posture (such as landscape / portrait switching, related games, magnetometer posture calibration), vibration recognition-related functions (such as pedometer, taps), etc. Other sensors that may be configured in the mobile phone, such as gyroscopes, barometers, hygrometers, thermometers, and infrared sensors, will not be described in detail here.

[0183] Audio circuit 960, speaker 961, and microphone 962 provide an audio interface between the user and the mobile phone. Audio circuit 960 converts received audio data into electrical signals and transmits them to speaker 961, where speaker 961 converts them into sound signals for output. On the other hand, microphone 962 converts collected sound signals into electrical signals, which are received by audio circuit 960, converted into audio data, and then processed by processor 980 before being transmitted via RF circuit 910 to, for example, another mobile phone, or the audio data can be output to memory 920 for further processing.

[0184] WiFi is a short-range wireless transmission technology. Mobile phones using the WiFi module 970 can help users send and receive emails, browse web pages, and access streaming media, providing users with wireless broadband internet access. Although Figure 9 The WiFi module 970 is shown, but it is understood that it is not an essential component of a mobile phone and can be omitted as needed without changing the essence of the invention.

[0185] The processor 980 is the control center of the mobile phone, connecting various parts of the phone through various interfaces and lines. It performs various functions and processes data by running or executing software programs and / or modules stored in the memory 920, and by calling data stored in the memory 920, thereby managing the phone as a whole. Optionally, the processor 980 may include one or more processing units; optionally, the processor 980 may integrate an application processor and a modem processor, wherein the application processor mainly handles the operating system, user interface, and applications, and the modem processor mainly handles wireless communication. It is understood that the aforementioned modem processor may also not be integrated into the processor 980.

[0186] The phone also includes a power supply 990 (such as a battery) that supplies power to various components. Optionally, the power supply can be logically connected to the processor 980 through a power management system, thereby enabling functions such as charging, discharging, and power consumption management through the power management system.

[0187] Although not shown, mobile phones may also include a camera, Bluetooth module, etc., which will not be described in detail here.

[0188] In this embodiment of the application, the processor 980 included in the terminal also has the function of performing the various steps of the page processing method described above.

[0189] This application also provides a server; please refer to [link / reference]. Figure 10 , Figure 10 This is a schematic diagram of a server structure provided in an embodiment of this application. The server 1000 can vary significantly due to different configurations or performance. It may include one or more central processing units (CPUs) 1022 (e.g., one or more processors) and memory 1032, and one or more storage media 1030 (e.g., one or more mass storage devices) for storing application programs 1042 or data 1044. The memory 1032 and storage media 1030 can be temporary or persistent storage. The program stored in the storage media 1030 may include one or more modules (not shown in the diagram), each module may include a series of instruction operations on the server. Furthermore, the CPU 1022 may be configured to communicate with the storage media 1030 and execute the series of instruction operations in the storage media 1030 on the server 1000.

[0190] Server 1000 may also include one or more power supplies 1026, one or more wired or wireless network interfaces 1050, one or more input / output interfaces 1058, and / or one or more operating systems 1041, such as Windows Server™, Mac OS X™, Unix™, Linux™, FreeBSD™, etc.

[0191] The steps performed by the management device in the above embodiments can be based on this Figure 10 The server structure shown.

[0192] This application also provides a computer-readable storage medium storing detection instructions for an object set, which, when executed on a computer, cause the computer to perform the aforementioned actions. Figures 3 to 7 The steps performed by the object set detection device in the method described in the illustrated embodiment.

[0193] This application also provides a computer program product that includes detection instructions for a set of objects, which, when run on a computer, causes the computer to perform the aforementioned... Figures 3 to 7 The steps performed by the object set detection device in the method described in the illustrated embodiment.

[0194] This application embodiment also provides a detection system for an object set, the object set detection system may include... Figure 8 The detection device for the set of objects in the described embodiments, or Figure 9 The terminal device in the described embodiments, or Figure 10 The server described.

[0195] In one possible scenario, the network resource management method of this application is applied to a blockchain device, i.e., the authoritative DNS, LDNS, or terminal is a blockchain device, and this blockchain device is a node in the blockchain. This is described below with reference to the accompanying drawings; see also... Figure 11A The data sharing system shown, data sharing system 1100, refers to a system for data sharing between nodes. This data sharing system may include multiple nodes 1101, which can refer to various clients within the data sharing system. Each node 1101, during normal operation, can receive input information and maintain shared data within the data sharing system based on the received input information. To ensure information interoperability within the data sharing system, information connections can exist between each node, allowing information transmission between nodes. For example, when any node in the data sharing system receives input information, other nodes in the system obtain this input information according to a consensus algorithm and store it as data in the shared data, ensuring consistency of data stored on all nodes in the data sharing system.

[0196] Each node in the data sharing system has a corresponding node identifier, and each node can also store the node identifiers of other nodes in the data sharing system. This allows for the subsequent broadcasting of generated blocks to other nodes in the data sharing system based on their node identifiers. Each node can maintain a node identifier list as shown in the table below, storing the node name and node identifier in this list. The node identifier can be an IP (Internet Protocol) address or any other information that can be used to identify the node. Table 1 only uses IP addresses as an example.

[0197] Table 1. Correspondence between node names and node identifiers

[0198] Node Name Node identifier Node 1 117.114.151.174 Node 2 117.116.189.145 … … Node N 119.123.789.258

[0199] Each node in the data-sharing system stores the same blockchain. A blockchain consists of multiple blocks; see [link to blockchain documentation]. Figure 11B A blockchain consists of multiple blocks. The genesis block includes a block header and a block body. The block header stores input information feature values, version number, timestamp, and difficulty value, while the block body stores the input information. The next block after the genesis block takes the genesis block as its parent block. The next block also includes a block header and a block body. The block header stores the input information feature values ​​of the current block, the block header feature values ​​of the parent block, version number, timestamp, and difficulty value, and so on. This ensures that the block data stored in each block is related to the block data stored in the parent block, guaranteeing the security of the input information in the blocks.

[0200] When generating the individual blocks in the blockchain, see Figure 11C When a node in the blockchain receives input information, it verifies the input information. After verification, it stores the input information in a memory pool and updates its hash tree used to record the input information. Then, it updates the timestamp to the time the input information was received and tries different random numbers multiple times to calculate the feature value, ensuring that the calculated feature value satisfies the following formula:

[0201] SHA256(SHA256(version+prev_hash+merkle_root+ntime+nbits+x))<TARGET

[0202] Wherein, SHA256 is the feature value algorithm used to calculate the feature value; version (version number) is the version information of the relevant block protocol in the blockchain; prev_hash is the block header feature value of the parent block of the current block; merkle_root is the feature value of the input information; ntime is the update time of the update timestamp; nbits is the current difficulty, which is a fixed value for a period of time and is determined again after exceeding the fixed time period; x is a random number; TARGET is the feature value threshold, which can be determined based on nbits.

[0203] Thus, when a random number satisfying the above formula is calculated, the information can be stored accordingly, generating a block header and a block body to obtain the current block. Subsequently, the node where the blockchain resides sends the newly generated block to other nodes in its data sharing system based on the node identifiers of other nodes in the data sharing system. The other nodes then verify the newly generated block and add it to their stored blockchain after verification.

[0204] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0205] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection between apparatuses or units through some interfaces, and may be electrical, mechanical, or other forms.

[0206] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0207] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0208] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, a detection device for a collection of objects, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0209] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A method for detecting a set of objects, characterized in that, include: Obtain the relationship data between each object in the set of objects to be processed. The relationship data includes interaction data and location data. The interaction data includes any one or a combination of team relationship, game relationship, communication relationship, kill relationship and support relationship. The location data includes any one or a combination of geographical relationship and server relationship. The relationship data is obtained by traversing the data based on the correlation between the interaction data and the location data. The relational data are linked using triples to obtain a relational chain; The relationship chains are integrated to determine the relationship graph network; The preset seed object is used to traverse the nodes of the relational graph network to obtain the diffusion object associated with the preset seed object; The nodes corresponding to the diffusion objects in the relational graph network are connected to obtain a subgraph network; The graph neural network is invoked to extract features from the subgraph network to obtain the target graph features; Based on the target graph features, a target graph network similar to the subgraph network is obtained by filtering in the relation graph network, and a set of target objects in the set of objects to be processed is determined according to the target graph network. This includes: traversing the relation graph network based on the target graph features to obtain a similarity parameter; filtering the target graph network similar to the subgraph network according to the similarity parameter; performing prediction weighting on the objects contained in the target graph network and the diffusion objects to obtain a merging suspicion parameter; and determining the set of target objects in the set of objects to be processed based on the merging suspicion parameter, wherein the object types in the set of target objects are similar to the types of the preset seed objects.

2. The method according to claim 1, characterized in that, The step of traversing the nodes of the relational graph network according to a preset seed object to obtain a diffusion object associated with the preset seed object includes: Based on the preset seed object, it roams through the nodes of the relationship graph network to determine various roaming relationships; The walk parameters corresponding to each walk relationship are calculated based on the restarted random walk algorithm until the walk parameters converge, and then the walk score vector corresponding to the walk relationship is determined. The correlation is calculated based on the wandering score vector to obtain the degree of suspicion; If the suspicion level reaches a preset threshold, then the node corresponding to the wandering relationship is determined to be a diffusion object associated with the preset seed object.

3. The method according to claim 2, characterized in that, The method further includes: Repeated walks are performed at the nodes of the relational graph network to obtain object update parameters; If the object update parameter indicates that the diffusion object has not increased, then the traversal of nodes in the relational graph network is stopped.

4. The method according to claim 1, characterized in that, The process of calling a graph neural network to extract features from the subgraph network to obtain target graph features includes: The preset seed object and the diffusion object are marked as black samples, and the objects in the relationship graph network other than the preset seed object and the diffusion object are marked as white samples; A preset neural network configured based on a nested formula of depth maps is invoked, and the black sample and the white sample are input into the preset neural network to obtain classification loss information; Based on the classification loss information, the preset neural network is trained using the backpropagation algorithm to obtain the graph neural network; The subgraph network is input into the graph neural network for feature extraction to obtain the target graph features.

5. The method according to any one of claims 1-4, characterized in that, The detection method for the object set is applied to a blockchain device, which is a node in the blockchain.

6. A device for detecting a set of objects, characterized in that, include: The acquisition unit is used to acquire the relationship data between each object in the set of objects to be processed. The relationship data includes interaction data and location data. The interaction data includes any one or a combination of team relationship, game relationship, communication relationship, kill relationship and support relationship. The location data includes any one or a combination of geographical relationship and server relationship. The unit traverses the data based on the correlation between the interaction data and the location data to obtain the relationship data. The relational data are linked using triples to obtain a relational chain; The relationship chains are integrated to determine the relationship graph network; The walking unit is used to walk through the nodes of the relational graph network according to the preset seed object in order to obtain the diffusion object associated with the preset seed object; A detection unit is used to connect the nodes corresponding to the diffusion object in the relational graph network to obtain a subgraph network; The detection unit is also used to call a graph neural network to extract features from the subgraph network in order to obtain the target graph features; The detection unit is further configured to: filter the relation graph network based on the target graph features to obtain a target graph network similar to the subgraph network; and determine the target object set in the set of objects to be processed based on the target graph network. This includes: traversing the relation graph network based on the target graph features to obtain a similarity parameter; filtering the target graph network based on the similarity parameter to obtain a target graph network similar to the subgraph network; performing prediction weighting on the objects contained in the target graph network and the diffusion object to obtain a merging suspicion parameter; and determining the target object set in the set of objects to be processed based on the merging suspicion parameter, wherein the object type in the target object set is similar to the type of the preset seed object.

7. The apparatus according to claim 6, characterized in that, The traveling unit is specifically used for: Based on the preset seed object, it roams through the nodes of the relationship graph network to determine various roaming relationships; The walk parameters corresponding to each walk relationship are calculated based on the restarted random walk algorithm until the walk parameters converge, and then the walk score vector corresponding to the walk relationship is determined. The correlation is calculated based on the wandering score vector to obtain the degree of suspicion; If the suspicion level reaches a preset threshold, then the node corresponding to the wandering relationship is determined to be a diffusion object associated with the preset seed object.

8. The apparatus according to claim 7, characterized in that, The traveling unit is specifically used for: Repeated walks are performed at the nodes of the relational graph network to obtain object update parameters; If the object update parameter indicates that the diffusion object has not increased, then the traversal of nodes in the relational graph network is stopped.

9. The apparatus according to claim 6, characterized in that, The detection unit is specifically used for: The preset seed object and the diffusion object are marked as black samples, and the objects in the relationship graph network other than the preset seed object and the diffusion object are marked as white samples; A preset neural network configured based on a nested formula of depth maps is invoked, and the black sample and the white sample are input into the preset neural network to obtain classification loss information; Based on the classification loss information, the preset neural network is trained using the backpropagation algorithm to obtain the graph neural network; The subgraph network is input into the graph neural network for feature extraction to obtain the target graph features.

10. The apparatus according to any one of claims 6-9, characterized in that, The detection method for the object set is applied to a blockchain device, which is a node in the blockchain.

11. A computer device, characterized in that, The computer device includes a processor and memory: The memory is used to store program code; the processor is used to execute the object set detection method according to any one of claims 1 to 5 according to the instructions in the program code.

12. A computer program product comprising a computer program / instructions, characterized in that, When the computer program / instructions are executed by the processor, they implement the steps of the object set detection method according to any one of claims 1 to 5.

13. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores instructions that, when executed on a computer, cause the computer to perform the method for detecting a set of objects as described in any one of claims 1 to 5.

Images