Game cheat detection method, device, electronic device and storage medium
By obtaining the resource interaction map of gamers and using graph node embedding algorithms and clustering algorithms, the problem of low accuracy of game plug-in detection in the existing technology is solved, and more efficient plug-in player recognition is achieved.
Patent Information
- Application Number
- CN202011612751.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2020-12-30
- Publication Date
- 2024-11-12
- Estimated Expiration
- 2040-12-30
AI Technical Summary
The existing game plug-in detection methods are relatively low in accuracy, and plug-in developers can quickly modify and hide client processes, resulting in inaccurate detection.
By obtaining the resource interaction map of gamers, using the graph node embedding algorithm and clustering algorithm to generate representation vectors, filter out suspected plug-in players, and combine resource interaction data and player behavior characteristics to improve the accuracy of detection.
It improves the accuracy of game plug-in detection, can more effectively identify and filter out suspected plug-in players, and enhances the reliability of detection.
Smart Images

Figure CN112587932B_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of game technology, and in particular to a method, device, electronic device, and storage medium for detecting game cheats. Background Technology
[0002] Game cheats generally refer to cheating programs or software that modify game data to gain an advantage for players. They use computer technology to perform non-original operations on one or more software programs, tampering with the game's original settings and rules, greatly enhancing the skills and extraordinary abilities of game characters, thereby easily obtaining the benefits of victory, rewards, and pleasure. Cheating programs created by altering parts of the software's program seriously undermine the fairness of the game.
[0003] Current methods for detecting game cheats primarily rely on the game process. They check whether the game client has enabled any cheat processes listed in the blacklist. If a blacklisted cheat process is detected, the client is considered to have installed a game cheat.
[0004] However, because existing detection methods are relatively simple, cheat developers can quickly modify and hide the client-side cheat process, resulting in low accuracy of existing methods for detecting game cheats. Summary of the Invention
[0005] The purpose of this application is to address the shortcomings of the prior art by providing a method, apparatus, electronic device, and storage medium for detecting game cheats, which can improve the accuracy of game cheat detection.
[0006] To achieve the above objectives, the technical solutions adopted in the embodiments of this application are as follows:
[0007] In a first aspect, embodiments of this application provide a method for detecting game cheats, including: acquiring a resource interaction graph based on a target game, wherein the resource interaction graph includes a graph constructed with multiple game players as nodes and the resource interaction data flow between the multiple game players as edges, and the multiple game players include at least one cheat player and multiple players to be tested;
[0008] Based on the preset graph node embedding algorithm, obtain the representation vector of each node in the resource interaction graph;
[0009] Based on the representation vector of each node, a pre-defined clustering algorithm is used to generate clustering results for similar nodes.
[0010] Based on the clustering results, suspected cheaters were identified from among multiple test players.
[0011] Optionally, obtain a resource interaction graph based on the target game, including:
[0012] Obtain the initial resource interaction data of each game player. The initial resource interaction data includes: the resource interaction initiator, the resource interaction receiver, and the resources interacted with.
[0013] Based on the initial resource interaction data and preset filtering conditions, target resource interaction data is filtered out. The target resource interaction data includes resource interaction data of at least one cheating player and resource interaction data of multiple players to be tested. The preset filtering conditions include the identifiers of known cheating players.
[0014] Based on the target resource interaction data, a resource interaction graph is constructed with multiple game players as nodes and the flow of resource interaction data between multiple game players as edges.
[0015] Optionally, based on the representation vectors of each node, a preset clustering algorithm is used to generate clustering results for similar nodes, including:
[0016] Based on the representation vector of each node, a pre-defined clustering algorithm is used to calculate the similarity between nodes;
[0017] Based on the similarity between nodes, multiple nodes are divided into multiple clusters, and each cluster includes at least a preset number of nodes.
[0018] Optionally, based on the group clustering results, suspected cheaters can be identified from among multiple test players, including:
[0019] Obtain the suspicion score for each cluster as a cheating cluster. The suspicion score is used to indicate the ratio of the number of players marked as cheating players in the cluster to the total number of players in the cluster.
[0020] Based on the suspicion level of each cluster as a cheating cluster and the preset suspicion threshold, suspected cheating players are selected from multiple players to be tested.
[0021] Optionally, based on a preset graph node embedding algorithm, the representation vector of each node in the resource interaction graph is obtained, including:
[0022] The graph embedding algorithm based on random walks extracts the contextual relationships of each node in the resource interaction graph. The contextual relationships of each node are used to indicate the upstream and downstream nodes of each node in the random walk path.
[0023] Based on the contextual relationships of each node in the resource interaction graph, the first representation vector of each node in the resource interaction graph is obtained. The first representation vector of each node is used to indicate the connection relationship between adjacent nodes.
[0024] Optionally, based on a preset graph node embedding algorithm, the representation vector of each node in the resource interaction graph is obtained, including:
[0025] Based on the deep autoencoder algorithm, the second representation vector of each node in the resource interaction graph is obtained. The second representation vector of each node is used to indicate the connection relationship between the node and its surrounding nodes.
[0026] Optionally, based on the representation vectors of each node, a preset clustering algorithm is used to generate clustering results for similar nodes, including:
[0027] Based on the first and / or second representation vectors of each node, a preset clustering algorithm is used to generate the first and / or second group clustering results of similar nodes.
[0028] Accordingly, based on the group clustering results, suspected cheaters were identified from multiple test players, including:
[0029] Based on the clustering results of the first group and / or the clustering results of the second group, suspected cheaters are screened out from multiple test players.
[0030] Optionally, based on the clustering results of the first group and / or the clustering results of the second group, suspected cheaters are screened from multiple test players, including:
[0031] Obtain the first likelihood that each cluster in the first group clustering result is an external cluster and / or the second likelihood that each cluster in the second group clustering result is an external cluster;
[0032] Based on the first suspicion level and / or the second suspicion level, as well as the preset suspicion threshold, suspected cheaters are selected from multiple test players.
[0033] Optionally, after generating the first group clustering results and / or the second group clustering results of similar nodes using a preset clustering algorithm, the method further includes:
[0034] The clustering results of the first group and / or the clustering results of the second group are displayed according to the preset display method.
[0035] Secondly, embodiments of this application provide a device for detecting game cheats, including: a first acquisition module, a second acquisition module, a generation module, and a filtering module;
[0036] The first acquisition module is used to acquire a resource interaction graph based on the target game. The resource interaction graph includes a graph constructed with multiple game players as nodes and the resource interaction data flow between multiple game players as edges. The multiple game players include at least one cheater and multiple players to be tested.
[0037] The second acquisition module is used to acquire the representation vector of each node in the resource interaction graph according to the preset graph node embedding algorithm.
[0038] The generation module is used to generate clustering results of similar nodes based on the representation vectors of each node using a preset clustering algorithm;
[0039] The filtering module is used to filter out suspected cheaters from multiple test players based on the results of group clustering.
[0040] Optionally, the first acquisition module is specifically used to acquire the initial resource interaction data of each game player. The initial resource interaction data includes: the resource interaction initiator, the resource interaction receiver, and the resources being interacted with.
[0041] Based on the initial resource interaction data and preset filtering conditions, target resource interaction data is filtered out. The target resource interaction data includes resource interaction data of at least one cheating player and resource interaction data of multiple players to be tested. The preset filtering conditions include the identifiers of known cheating players.
[0042] Based on the target resource interaction data, a resource interaction graph is constructed with multiple game players as nodes and the flow of resource interaction data between multiple game players as edges.
[0043] Optionally, the generation module is specifically used to calculate the similarity between nodes based on the representation vector of each node using a preset clustering algorithm;
[0044] Based on the similarity between nodes, multiple nodes are divided into multiple clusters, and each cluster includes at least a preset number of nodes.
[0045] Optionally, the filtering module is specifically used to obtain the suspicion degree of each cluster as a cheating cluster. The suspicion degree is used to indicate the ratio of the number of players marked as cheating players in the cluster to the total number of players in the cluster.
[0046] Based on the suspicion level of each cluster as a cheating cluster and the preset suspicion threshold, suspected cheating players are selected from multiple players to be tested.
[0047] Optionally, the second acquisition module is specifically used to extract the contextual relationships of each node in the resource interaction graph using a graph embedding algorithm based on random walk. The contextual relationships of each node are used to indicate the upstream and downstream nodes of each node in the random walk path.
[0048] Based on the contextual relationships of each node in the resource interaction graph, the first representation vector of each node in the resource interaction graph is obtained. The first representation vector of each node is used to indicate the connection relationship between adjacent nodes.
[0049] Optionally, the second acquisition module is specifically used to acquire the second representation vector of each node in the resource interaction graph according to the deep autoencoder algorithm. The second representation vector of each node is used to indicate the connection relationship between the node and its surrounding nodes.
[0050] Optionally, the generation module is specifically used to generate a first group clustering result and / or a second group clustering result of similar nodes based on the first representation vector and / or the second representation vector of each node using a preset clustering algorithm; the filtering module is specifically used to filter out suspected cheating players from multiple players to be tested based on the first group clustering result and / or the second group clustering result.
[0051] Optionally, the filtering module is specifically used to obtain the first likelihood that each cluster in the first group clustering result is an external cluster and / or the second likelihood that each cluster in the second group clustering result is an external cluster;
[0052] Based on the first suspicion level and / or the second suspicion level, as well as the preset suspicion threshold, suspected cheaters are selected from multiple test players.
[0053] Optionally, the above device further includes: a display module, used to display the clustering results of the first group and / or the clustering results of the second group according to a preset display mode.
[0054] Thirdly, embodiments of this application provide an electronic device, including: a processor, a storage medium, and a bus. The storage medium stores machine-readable instructions executable by the processor. When the electronic device is running, the processor communicates with the storage medium via the bus, and the processor executes the machine-readable instructions to perform the steps of the game cheat detection method described in the first aspect.
[0055] Fourthly, embodiments of this application provide a storage medium storing a computer program, which, when run by a processor, executes the steps of the game cheat detection method described in the first aspect.
[0056] The beneficial effects of this application are:
[0057] The game cheat detection method, apparatus, electronic device, and storage medium provided in this application include: acquiring a resource interaction graph based on a target game, wherein the resource interaction graph is a graph constructed with multiple game players as nodes and the resource interaction data flow between the multiple game players as edges, and the multiple game players include at least one cheater and multiple players to be tested; acquiring the representation vector of each node in the resource interaction graph according to a preset graph node embedding algorithm; generating a group clustering result of similar nodes using a preset clustering algorithm based on the representation vector of each node; and filtering out suspected cheaters from among the multiple players to be tested based on the group clustering result. In this detection process, because the detection based on the resource interaction graph fully considers the attributes of the nodes themselves and the structural attributes of the resource interaction graph, the accuracy of filtering out suspected cheaters from among the multiple players to be tested based on the cheaters in the resource interaction graph can be improved. Attached Figure Description
[0058] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.
[0059] Figure 1 A cloud interaction system architecture diagram provided for an embodiment of this application;
[0060] Figure 2 A flowchart illustrating a method for detecting game cheats provided in this application embodiment;
[0061] Figure 3 A flowchart illustrating another method for detecting game cheats provided in this application embodiment;
[0062] Figure 4 A flowchart illustrating another method for detecting game cheats provided in this application embodiment;
[0063] Figure 5 A flowchart illustrating another method for detecting game cheats provided in this application embodiment;
[0064] Figure 6 A flowchart illustrating another method for detecting game cheats provided in this application embodiment;
[0065] Figure 7 A flowchart illustrating another method for detecting game cheats provided in this application embodiment;
[0066] Figure 8 A flowchart illustrating another method for detecting game cheats provided in this application embodiment;
[0067] Figure 9 A flowchart illustrating another method for detecting game cheats provided in this application embodiment;
[0068] Figure 10 A schematic diagram of the functional modules of a game cheat detection device provided in an embodiment of this application;
[0069] Figure 11 A schematic diagram of the functional modules of another game cheat detection device provided in an embodiment of this application;
[0070] Figure 12 This is a schematic diagram of an electronic device structure provided in an embodiment of this application. Detailed Implementation
[0071] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.
[0072] Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of the application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.
[0073] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.
[0074] There are many types of game cheats currently available. For example, there are auto-cheat game cheats. Auto-cheat game cheats can use some program scripts to make players automatically perform tasks, kill monsters, level up, and earn money, and transfer and collect money, which seriously undermines the fairness of the game.
[0075] Existing methods for detecting game cheats include: detecting cheat processes running in the game client; statistically analyzing player behavior based on experiential features and then investigating players accordingly; identifying cheat software windows on the client screen; and using traditional machine learning methods. However, because these methods are relatively simple, they suffer from low accuracy.
[0076] In view of this, embodiments of this application provide a method for detecting game cheats. This method can perform detection based on a resource interaction graph. In this process, the attributes of the nodes themselves and the structural attributes of the resource interaction graph are fully considered, so that when filtering suspected cheat players from multiple test players based on cheat players in the resource interaction graph, the detection accuracy can be improved.
[0077] In one embodiment of this application, the game cheat detection method can run on a terminal device or a server. The terminal device can be a local terminal device. When the game cheat detection method runs on a server, it can be implemented and executed based on a cloud interaction system, which can include a server and client devices.
[0078] Figure 1 A cloud interaction system architecture diagram is provided for embodiments of this application, such as... Figure 1 As shown, the system may include: a client device 10 and a server 20, wherein the client device 10 can be connected to the server 20 via a network 30.
[0079] In an optional implementation, various cloud applications, such as cloud gaming, can run under the cloud interaction system. Taking cloud gaming as an example, cloud gaming refers to a gaming method based on cloud computing. In the cloud gaming operation mode, the game program and the game screen presentation are separated. The storage and execution of the game character's transmission method are completed on the cloud gaming server. The client device 10 is used for receiving and sending data and presenting the game screen. For example, the client device 10 can be a display device with data transmission capabilities close to the user, such as a mobile terminal, television, computer, or PDA; however, the terminal device for information processing is the cloud gaming server in the cloud. When playing the game, the player operates the client device 10 to send operation commands to the cloud gaming server. The cloud gaming server runs the game according to the operation commands, encodes and compresses the game screen and other data, returns it to the client device 10 through the network, and finally, the client device 10 decodes and outputs the game screen.
[0080] In an alternative implementation, the terminal device can be a local terminal device. Taking a game as an example, the local terminal device stores the game program and is used to display the game screen. The local terminal device is used to interact with the player through a graphical user interface, that is, conventionally downloading, installing, and running the game program via an electronic device. The local terminal device can provide the graphical user interface to the player in various ways, such as rendering it on the terminal's display screen, or providing it to the player through holographic projection. For example, the local terminal device can include a display screen for displaying the graphical user interface, which includes game screens, and a processor for running the game, generating the graphical user interface, and controlling the display of the graphical user interface on the display screen.
[0081] Figure 2 This is a flowchart illustrating a method for detecting game cheats provided in an embodiment of this application. The execution subject of this method can be a computer, server, processor, mobile terminal, etc., and is not limited thereto. Figure 2 As shown, the detection method may include:
[0082] S101. Obtain the resource interaction graph based on the target game. The resource interaction graph includes a graph constructed with multiple game players as nodes and the resource interaction data flow between multiple game players as edges. The multiple game players include at least one cheater and multiple players to be tested.
[0083] The target game can contain multiple players, such as in multiplayer shooting games, multiplayer card games, and massively multiplayer online role-playing games (MMORPGs), without limitation. During gameplay, these players can interact with resources, generating target resource interaction data. Based on this target resource interaction data, a graph can be constructed using the players as nodes and the flow of resource interaction data between them as edges, resulting in a resource interaction graph. It should be noted that these multiple players can include at least one cheater and multiple test players. A cheater is a player exhibiting cheating behavior, and test players are those used to detect whether a player is suspected of cheating.
[0084] S102. Obtain the representation vector of each node in the resource interaction graph according to the preset graph node embedding algorithm.
[0085] A pre-defined graph node embedding algorithm is used to obtain the representation vectors of each node in the resource interaction graph. Optionally, this pre-defined graph node embedding algorithm may include, but is not limited to, vertex embedding methods, Structural Deep Network Embedding (SDNE) methods, etc., and is not limited here. One or more graph node embedding algorithms can be selected to obtain the representation vectors of each node in the resource interaction graph according to the actual application scenario, and is not limited here. The representation vector of a node can represent the node's own attributes and the structural attributes of the resource interaction graph.
[0086] S103. Based on the representation vector of each node, a preset clustering algorithm is used to generate the clustering results of similar nodes.
[0087] The preset clustering algorithm may include, but is not limited to, K-means clustering, Expectation-Maximum (EM) clustering with Gaussian Mixture Model (GMM), and density-based clustering algorithms, etc., without limitation. Density-based clustering algorithms may include density-based clustering of applications with noise (DBSCAN) and mean-shift clustering, without limitation; a suitable preset clustering algorithm can be selected based on the actual application scenario. After obtaining the representation vector of each node, the preset clustering algorithm can be used to generate a group clustering result of similar nodes based on the similarity of the nodes in their representation vectors. This group clustering result can represent the similarity between nodes.
[0088] S104. Based on the group clustering results, identify suspected cheaters from among multiple test players.
[0089] Since the resource interaction graph includes at least one cheater, and the group clustering result can represent the similarity between nodes, the game players with similar representation vectors to at least one cheater can be selected from multiple test players as suspected cheaters based on the group clustering result, thus realizing the detection of suspected cheaters.
[0090] In summary, the game cheat detection method provided in this application includes: acquiring a resource interaction graph based on the target game, wherein the resource interaction graph is constructed with multiple game players as nodes and the resource interaction data flow between the multiple game players as edges, and the multiple game players include at least one cheater and multiple players to be tested; acquiring the representation vector of each node in the resource interaction graph according to a preset graph node embedding algorithm; generating a group clustering result of similar nodes using a preset clustering algorithm based on the representation vector of each node; and filtering out suspected cheaters from the multiple players to be tested based on the group clustering result. In this detection process, because the detection based on the resource interaction graph fully considers the attributes of the nodes themselves and the structural attributes of the resource interaction graph, the accuracy of filtering out suspected cheaters from the multiple players to be tested based on the cheaters in the resource interaction graph can be improved.
[0091] Figure 3 This is a flowchart illustrating another method for detecting game cheats provided in an embodiment of this application. Optionally, as... Figure 3 As shown, the above-mentioned acquisition of the resource interaction graph based on the target game includes:
[0092] S201. Obtain the initial resource interaction data of each game player. The initial resource interaction data includes: the resource interaction initiator, the resource interaction receiver, and the resources involved in the interaction.
[0093] In this game, players can interact with each other by logging into the game client. During these interactions, initial resource exchange data is generated. This data may include, but is not limited to, the resource exchange initiator, the resource exchange recipient, and the exchanged resources. These resources may include, but are not limited to, virtual currency, virtual items, and game merchandise. For example, the resource exchange initiator and recipient can exchange resources. For the initiator, the exchanged resource could be game equipment; conversely, for the recipient, the exchanged resource could be virtual currency.
[0094] In some embodiments, the initial resource interaction data may also include the interaction time, that is, the time during which the resource interaction initiator and the resource interaction receiver engage in resource interaction.
[0095] S202. Based on the initial resource interaction data and preset filtering conditions, filter out the target resource interaction data. The target resource interaction data includes the resource interaction data of at least one cheating player and the resource interaction data of multiple players to be tested.
[0096] The preset filtering criteria include the identifiers of known cheating players. These criteria are used to filter target resource interaction data from the initial resource interaction data. Specifically, the preset filtering criteria can include the identifiers of known cheating players and the filtering conditions for the interaction resources. Based on the identifiers of known cheating players, resource interaction data of known cheating players can be filtered from the initial resource interaction data; based on the filtering conditions for the interaction resources, resource interaction data of multiple players to be tested can be filtered from the initial resource interaction data. It can be understood that through the above filtering, resource interaction data of suspected cheating players can be initially filtered from the initial resource interaction data to obtain the target resource interaction data. This improves the efficiency of subsequent filtering of suspected cheating players based on the target resource interaction data.
[0097] In some embodiments, the filtering criteria for interactive resources may include: the return rate threshold of the resource interaction initiator and the loss rate threshold of the resource interaction recipient. Optionally, the return rate and loss rate of the two players in the transaction can be obtained, and the return rate and loss rate can be compared with the return rate threshold and loss rate threshold, respectively. Based on the comparison results, resource interaction data exceeding these thresholds can be filtered out from the initial resource interaction data as target resource interaction data, so as to initially filter out resource interaction data of players who are not suspected of using cheats. However, the actual filtering criteria are not limited to this.
[0098] For example, if the preset return rate threshold is A and the preset loss rate threshold is B, and player A gives player B a piece of equipment, which is estimated to be worth W based on historical transaction prices, while player B gives player A in-game currency V, then player A's return rate is (VW) / W, and player B's loss rate is (WV) / V. Based on the return rate thresholds A and (VW) / W, and the loss rate thresholds B and (WV) / V, resource interaction data of players suspected of using cheats can be initially filtered from the initial resource interaction data. Of course, this application does not limit the specific filtering method here. It can be that both the return rate and the loss rate exceed the corresponding thresholds for filtering, or it can be that one of them exceeds the corresponding threshold for filtering.
[0099] S203. Based on the target resource interaction data, construct a resource interaction graph with multiple game players as nodes and the resource interaction data flow between multiple game players as edges.
[0100] Based on the above, after obtaining the target resource interaction data, a resource interaction graph can be constructed using multiple game players in the target resource interaction data as nodes and the flow of resource interaction data between multiple game players as edges.
[0101] In some embodiments, considering the trading characteristics of cheating players, for example, the money accumulation process of cheating players often has the following characteristics: cheating players will use a small number of traded items to cover up the transfer behavior, but these transactions are often a one-sided profit process; cheating players often concentrate virtual currency from a large number of cheating player accounts to a small number of cheating player accounts. Based on this trading characteristic, the flow of resource interaction data can be the flow of virtual currency in the transactions. If there are multiple transactions between two game players, then there can be multiple edges between these two game players in the constructed resource interaction graph, and each edge can be regarded as a transaction.
[0102] Figure 4 This is a flowchart illustrating another method for detecting game cheats provided in this application. Optionally, as... Figure 4 As shown, the above-mentioned clustering results of similar nodes are generated using a preset clustering algorithm based on the representation vectors of each node, including:
[0103] S301. Based on the representation vector of each node, calculate the similarity between each node using a preset clustering algorithm.
[0104] S302. Based on the similarity between nodes, divide multiple nodes into multiple clusters, and each cluster includes at least a preset number of nodes.
[0105] After obtaining the representation vector of each node, a preset clustering algorithm can be used to calculate the similarity between nodes based on the representation vector of each node. Multiple nodes are divided into multiple clusters according to the similarity. The multiple clusters are then filtered out according to the number of nodes included in each cluster, so that each filtered cluster can include a preset number of nodes.
[0106] In some embodiments, the DBSCAN clustering method can be used to generate clustering results for similar nodes. DBSCAN is a density-based algorithm based on high-density connected regions. It calculates the distance between each pair of nodes, dividing points that meet a preset distance threshold into clusters. The size of each cluster determines whether to retain it; points that cannot form clusters are considered discrete points. Optionally, each cluster can include at least 20 nodes. However, this application does not limit the neighborhood size; appropriate embedding algorithms can be selected based on the actual application scenario.
[0107] Figure 5This is a flowchart illustrating another method for detecting game cheats provided in an embodiment of this application. Optionally, as... Figure 5 As shown above, based on the group clustering results, suspected cheaters were identified from multiple test players, including:
[0108] S401. Obtain the suspicion level of each cluster as a cheating cluster. The suspicion level is used to indicate the ratio of the number of players marked as cheating players in the cluster to the total number of players in the cluster.
[0109] S402. Based on the suspicion level of each cluster as a cheating cluster and the preset suspicion threshold, filter out suspected cheating players from multiple players to be tested.
[0110] In this process, after obtaining multiple clusters based on a preset clustering algorithm, the number of players marked as cheaters within each cluster can be obtained and the total number of players within the cluster can be calculated. The ratio between the two can be recorded as the degree of suspicion that the cluster is a cheater cluster. Optionally, the degree of suspicion that each cluster is a cheater cluster can be compared with a preset suspicion threshold. If the degree of suspicion that a cluster is a cheater cluster is greater than the preset suspicion threshold, then the cluster can be considered a suspected cluster. Based on the suspected cluster, suspected cheaters can be screened out from multiple players to be tested.
[0111] Figure 6 This is a flowchart illustrating another method for detecting game cheats provided in this application. Optionally, as... Figure 6 As shown, the above-mentioned method, based on a preset graph node embedding algorithm, obtains the representation vector of each node in the resource interaction graph, including:
[0112] S501. A graph embedding algorithm based on random walks extracts the contextual relationships of each node in the resource interaction graph. The contextual relationships of each node are used to indicate the upstream and downstream nodes of each node in the random walk path.
[0113] S502. Based on the contextual relationship of each node in the resource interaction graph, obtain the first representation vector of each node in the resource interaction graph. The first representation vector of each node is used to indicate the connection relationship between adjacent nodes.
[0114] In some embodiments, a graph embedding algorithm based on random walks, such as the Node2vec vertex embedding method, can be used to obtain the representation vectors of each node in the resource interaction graph. These representation vectors can reflect the structural attributes of the nodes in the resource interaction graph. Node2vec is a graph embedding algorithm based on random walks and Skipgrams. Applying Node2vec allows the extraction of contextual relationships between nodes from the resource interaction graph through random walks. Then, based on these contextual relationships, the Skipgram algorithm is used to learn the relationships between nodes. The contextual relationships of each node can indicate its upstream and downstream nodes in the random walk path. By learning the relationships between nodes, the edge relationships between adjacent nodes in the resource interaction graph can be learned. The Skipgram method can predict the probability of context nodes using the central node. By sampling and traversing nodes, the interdependence and co-occurrence relationships among all nodes in the resource interaction graph can be learned.
[0115] In the random walk process, the Node2vec graph embedding algorithm can use a first parameter and a second parameter to control the probability of depth-first and breadth-first traversals. The first parameter increases the probability of a node returning to its previous node, allowing the node to prioritize breadth-first traversal. The second parameter increases the probability of a node reaching deeper nodes, allowing the node to prioritize depth-first traversal. It can be seen that, in this process, this application can control the path obtained by the walk sampling by controlling the magnitude of the first parameter. During the walk, the first parameter controls the probability of a node returning to its previous node. Increasing the first parameter makes the path obtained by the walk sampling more inclined to traverse the surrounding area, which is more in line with the characteristic that the relationships between game players in the resource interaction graph are concentrated near the nodes. This affects the vector generated by the node representation, so that the first representation vector of each node can better express the similarity between nodes in the same group. Subsequently, based on the first representation vector, a preset clustering algorithm is used to generate the group clustering results of similar nodes. Based on the group clustering results, the accuracy of screening suspected cheaters can be improved when screening multiple test players.
[0116] Optionally, the above-mentioned method of obtaining the representation vector of each node in the resource interaction graph according to a preset graph node embedding algorithm includes:
[0117] Based on the deep autoencoder algorithm, the second representation vector of each node in the resource interaction graph is obtained. The second representation vector of each node is used to indicate the connection relationship between the node and its surrounding nodes.
[0118] The main idea of the deep autoencoder algorithm is to compress the input data into a low-dimensional vector, and then use this encoding to reconstruct the input to obtain the output. By reducing the difference between the input and output, information loss can be minimized. Since the cheating community is concentrated in transactions, the neighboring nodes of cheating nodes may also be cheating nodes. Optionally, this deep autoencoder algorithm can be the Structural Deep Network Embedding (SDNE) method. SDNE is a graph embedding algorithm based on first-order and second-order estimation of graph nodes. The input of the SDNE algorithm can be all the neighboring nodes of each node in the resource interaction graph, i.e., the vector composed of each row of the network adjacency matrix. The output second representation vector can reflect the neighbor similarity of each node, so that the second representation vector obtained based on the SDNE algorithm can make nodes with high similarity in the graph also have high similarity in the low-dimensional representation vector.
[0119] In some embodiments, the SDNE model can be trained based on a loss function. By determining whether the function value of the loss function meets a preset threshold, the network parameters in the SDNE model can be adjusted. This allows the second representation vector of each node in the resource interaction graph to be obtained based on the trained SDNE model, so that nodes with similar adjacency information have closer distances in the representation vector. Optionally, the loss function L... mix This can include: the loss function L of the first-order relation 1st Loss function L of second-order relation 2nd and the regularized loss function L reg This expression can be written as L mix =L 2nd +αL 1st +νL reg Where α and v can be user-defined hyperparameters used to control L 1st L 2nd and L reg The relative size of the nodes. First-order relationships between nodes refer to two nodes directly connected by an edge. In this case, the two nodes are generally similar, and the loss function for first-order relationships allows nodes with first-order relationships to express themselves as closely as possible. Second-order relationships between nodes refer to two nodes that are simultaneously connected to one or more of the same nodes. In this case, the two nodes are also similar, and the similarity between the two nodes depends on how many nodes are simultaneously connected to them. The loss function for second-order relationships allows nodes with second-order relationships to express themselves as closely as possible. The regularization loss function is the second norm of the parameters in the entire network structure, used to avoid overfitting in the SDNE model. It can be trained using the stochastic gradient descent (backpropagation) algorithm.
[0120] Figure 7 This is a flowchart illustrating another method for detecting game cheats provided in an embodiment of this application. Optionally, as... Figure 7 As shown, the above-mentioned clustering results of similar nodes are generated using a preset clustering algorithm based on the representation vectors of each node, including:
[0121] S601. Based on the first representation vector and / or the second representation vector of each node, a preset clustering algorithm is used to generate the first group clustering result and / or the second group clustering result of similar nodes.
[0122] Specifically, after obtaining the first representation vector of each node, a first pre-set clustering algorithm is used to generate a first group clustering result of similar nodes based on the first representation vector of each node; and a second pre-set clustering algorithm is used to generate a second group clustering result of similar nodes based on the second representation vector of each node. Optionally, the first pre-set clustering algorithm and the second pre-set clustering algorithm can be the same or different, which is not limited here.
[0123] Accordingly, based on the group clustering results, suspected cheaters were identified from multiple test players, including:
[0124] S602. Based on the clustering results of the first group and / or the clustering results of the second group, screen out suspected cheaters from multiple test players.
[0125] Specifically, if a first group clustering result is obtained, suspected cheaters can be identified from multiple test players based on this result; similarly, if a second group clustering result is obtained, suspected cheaters can be identified from multiple test players based on this result. In essence, if both first and second group clustering results are obtained, suspected cheaters can be identified by combining these results from multiple test players.
[0126] For example, if the first clustering result determines that player A is a suspected cheater, and the second clustering result also determines that player A is a suspected cheater, then it can be considered that player A has a high probability of being a suspected cheater. Alternatively, if the first clustering result determines that player B is not a suspected cheater, and the second clustering result also determines that player B is not a suspected cheater, then it can be considered that player B is not a suspected cheater. Of course, depending on the actual application scenario, the actual determination method is not limited to this.
[0127] Figure 8 This is a flowchart illustrating another method for detecting game cheats provided in this application. Optionally, as... Figure 8As shown, based on the clustering results of the first group and / or the second group, suspected cheaters were identified from multiple test players, including:
[0128] S701. Obtain the first likelihood that each cluster in the first group clustering result is an external cluster and / or the second likelihood that each cluster in the second group clustering result is an external cluster.
[0129] S702. Based on the first suspicion level and / or the second suspicion level and the preset suspicion threshold, select suspected cheaters from multiple test players.
[0130] For the first group clustering results, the first suspicion degree of each cluster being a cheating cluster can be obtained. In some embodiments, the first suspicion degree can be obtained by calculating the ratio of the number of cheating players in that cluster to the total number of game players. It is understood that the higher the ratio, the higher the first suspicion degree, and the higher the probability that the cluster is a cheating cluster. Of course, the method of obtaining the first suspicion degree is not limited to this. For example, the similarity between the cluster information and preset cheating cluster information can be obtained, and the first suspicion degree of the cluster being a cheating cluster can be determined based on the similarity. After obtaining the first suspicion degree, suspected cheating players can be screened from multiple test players based on the first suspicion degree and a preset suspicion threshold. If the first suspicion degree is greater than the preset suspicion threshold, the cluster corresponding to the first suspicion degree can be considered a cheating cluster, and then suspected cheating players can be screened from multiple test players based on the game players included in that cluster.
[0131] For the second group clustering results, the same screening method as the first group clustering results can be used to screen out suspected cheaters from multiple test players, which will not be elaborated here.
[0132] In some embodiments, to improve the accuracy of identifying suspected cheaters, suspected cheaters can be identified from multiple test players based on a first suspicion level, a second suspicion level, and a preset suspicion threshold. Optionally, if both the first suspicion level and the second suspicion level are greater than the preset suspicion threshold, test players belonging to both the first and second suspicion level clusters can be considered suspected cheaters. In this process, since the two representation vectors corresponding to the same node can be used to determine whether a node is a cheat node, i.e., whether a test player is a cheater, the advantages and disadvantages of using different preset graph node embedding algorithms to obtain the representation vectors of each node are fully considered. Therefore, the accuracy of identifying suspected cheaters can be improved.
[0133] Figure 9 This is a flowchart illustrating another method for detecting game cheats provided in an embodiment of this application. Optionally, as... Figure 8As shown, after generating the first group clustering results and / or the second group clustering results of similar nodes using a preset clustering algorithm, the above method further includes:
[0134] S801. Display the clustering results of the first group and / or the clustering results of the second group according to the preset display method.
[0135] In order to visualize the clustering results of the first group and / or the second group, so that users can intuitively understand the clustering results, preset display methods can be used, such as pie charts, line charts, bar charts, lists, etc., to visualize the clustering results of the first group and / or the second group. Taking a pie chart as an example, the pie chart can include multiple sector areas, each sector corresponding to a cluster. The size of the sector area represents the number of players in that cluster; a larger sector area indicates a larger number of players in the cluster. Furthermore, in some embodiments, the probability of each cluster being a cheating cluster can also be reflected in the pie chart. For example, the coloring of the sector area can reflect the probability that the cluster is a cheating cluster. The darker the color of the sector area, the higher the probability that the cluster is a cheating cluster (e.g., dark blue indicates a higher probability), and the lighter the color, the lower the probability (e.g., light blue indicates a lower probability). This allows users to intuitively understand the probability of each cluster being a cheating cluster, improving the user experience.
[0136] After identifying suspected cheaters, in some embodiments, these suspected cheaters can be used as new cheater samples and added to the resource transaction graph as new samples, thereby improving the reliability of the suspicion analysis for each cluster. Additionally, it is understood that these new cheater samples can also be used in other training scenarios to improve the efficiency of obtaining training samples.
[0137] Figure 10 This is a functional module diagram of a game cheat detection device provided in this application embodiment. The basic principle and technical effects of this device are the same as those of the corresponding method embodiment described above. For the sake of brevity, parts not mentioned in this embodiment can be referred to the corresponding content in the method embodiment. Figure 10 As shown, the detection device 100 includes: a first acquisition module 110, a second acquisition module 120, a generation module 130, and a screening module 140;
[0138] The first acquisition module 110 is used to acquire a resource interaction graph based on the target game. The resource interaction graph includes a graph constructed with multiple game players as nodes and the resource interaction data flow between the multiple game players as edges. The multiple game players include at least one cheating player and multiple players to be tested.
[0139] The second acquisition module 120 is used to acquire the representation vector of each node in the resource interaction graph according to the preset graph node embedding algorithm.
[0140] The generation module 130 is used to generate a group clustering result of similar nodes based on the representation vector of each node using a preset clustering algorithm;
[0141] The filtering module 140 is used to filter out suspected cheaters from multiple test players based on the group clustering results.
[0142] Optionally, the first acquisition module 110 is specifically used to acquire the initial resource interaction data of each game player. The initial resource interaction data includes: the resource interaction initiator, the resource interaction receiver, and the resources interacted with. Based on the initial resource interaction data and preset filtering conditions, target resource interaction data is filtered out. The target resource interaction data includes the resource interaction data of at least one cheating player and the resource interaction data of multiple players to be tested. The preset filtering conditions include the identifiers of known cheating players. Based on the target resource interaction data, a resource interaction graph is constructed with multiple game players as nodes and the flow of resource interaction data between multiple game players as edges.
[0143] Optionally, the generation module 130 is specifically used to calculate the similarity between nodes using a preset clustering algorithm based on the representation vector of each node; and to divide multiple nodes into multiple clusters based on the similarity between nodes, with each cluster including at least a preset number of nodes.
[0144] Optionally, the filtering module 140 is specifically used to obtain the suspicion degree of each cluster as a cheating cluster. The suspicion degree is used to indicate the ratio of the number of players marked as cheating players in the cluster to the total number of game players in the cluster. Based on the suspicion degree of each cluster as a cheating cluster and the preset suspicion threshold, suspected cheating players are filtered out from multiple players to be tested.
[0145] Optionally, the second acquisition module 120 is specifically used to extract the context relationship of each node in the resource interaction graph based on the random walk graph embedding algorithm. The context relationship of each node is used to indicate the upstream and downstream nodes of each node in the random walk path. Based on the context relationship of each node in the resource interaction graph, the first representation vector of each node in the resource interaction graph is obtained. The first representation vector of each node is used to indicate the connection relationship between adjacent nodes.
[0146] Optionally, the second acquisition module 120 is specifically used to acquire the second representation vector of each node in the resource interaction graph according to the deep autoencoder algorithm. The second representation vector of each node is used to indicate the connection relationship between the node and its surrounding nodes.
[0147] Optionally, the generation module 130 is specifically used to generate a first group clustering result and / or a second group clustering result of similar nodes based on the first representation vector and / or the second representation vector of each node using a preset clustering algorithm;
[0148] The filtering module 140 is specifically used to filter out suspected cheaters from multiple test players based on the clustering results of the first group and / or the clustering results of the second group.
[0149] Optionally, the filtering module 140 is specifically used to obtain the first suspicion degree that each cluster in the first group clustering result is a cheating cluster and / or the second suspicion degree that each cluster in the second group clustering result is a cheating cluster; based on the first suspicion degree and / or the second suspicion degree and the preset suspicion threshold, suspected cheating players are filtered out from multiple players to be tested.
[0150] Figure 11 This is a schematic diagram of the functional modules of another game cheat detection device provided in an embodiment of this application. Optionally, as... Figure 11 As shown, the detection device 100 further includes a display module 150, used to display the clustering results of the first group and / or the clustering results of the second group according to a preset display mode.
[0151] The above-described device is used to execute the method provided in the foregoing embodiments, and its implementation principle and technical effect are similar, so they will not be described again here.
[0152] These modules can be one or more integrated circuits configured to implement the above methods, such as one or more Application Specific Integrated Circuits (ASICs), one or more Digital Signal Processors (DSPs), or one or more Field Programmable Gate Arrays (FPGAs). Alternatively, when a module is implemented using processing element scheduler code, the processing element can be a general-purpose processor, such as a Central Processing Unit (CPU) or other processor capable of calling program code. Furthermore, these modules can be integrated together as a system-on-a-chip (SoC).
[0153] Figure 12 This is a schematic diagram of an electronic device structure provided as an embodiment of this application. Figure 12As shown, the electronic device may include a processor 210, a storage medium 220, and a bus 230. The storage medium 220 stores machine-readable instructions executable by the processor 210. When the electronic device is running, the processor 210 communicates with the storage medium 220 via the bus 230, and the processor 210 executes the machine-readable instructions to perform the steps of the above method embodiment. The specific implementation and technical effects are similar and will not be described in detail here.
[0154] Optionally, this application also provides a storage medium storing a computer program, which, when run by a processor, executes the steps of the above-described method embodiments. The specific implementation and technical effects are similar and will not be repeated here.
[0155] In the several embodiments provided in this application, it should be understood that the disclosed apparatus 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 through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.
[0156] 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.
[0157] 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 in a combination of hardware and software functional units.
[0158] The integrated units implemented as software functional units described above can be stored in a computer-readable storage medium. These software functional units, stored in a storage medium, include several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute some steps of the methods of 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.
[0159] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes the element.
[0160] The above are merely preferred embodiments of this application and are not intended to limit this application. Various modifications and variations are possible for those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application. It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need further definition and explanation in subsequent figures. The above are merely preferred embodiments of this application and are not intended to limit this application. Various modifications and variations are possible for those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.
Claims
1. A method for detecting game cheats, characterized in that: include: Obtaining a resource interaction graph based on a target game, wherein the resource interaction graph includes a graph constructed with multiple game players as nodes and resource interaction data flows between the multiple game players as edges, wherein the multiple game players include at least one cheat player and multiple players to be tested; According to a preset graph node embedding algorithm, a representation vector of each node in the resource interaction graph is obtained; According to the representation vector of each node, a preset clustering algorithm is used to generate a group clustering result of similar nodes; According to the group clustering result, screening out suspected cheating players from the plurality of players to be tested; The method of generating a group clustering result of similar nodes by using a preset clustering algorithm according to the representation vector of each node includes: Calculate the similarity between the nodes using a preset clustering algorithm according to the first characterization vector and / or the second characterization vector of each node, wherein the first characterization vector of each node is used to indicate the edge relationship between adjacent nodes, and the second characterization vector of each node is used to indicate the edge relationship between the node and surrounding nodes; According to the similarity between the nodes, the plurality of nodes are divided into a plurality of clusters, each of the clusters including at least a preset number of nodes.
2. The method according to claim 1, characterized in that The obtaining of a resource interaction graph based on a target game includes: Acquire initial resource interaction data of each of the game players, the initial resource interaction data including: a resource interaction initiator, a resource interaction receiver, and interacted resources; Filtering out target resource interaction data according to the initial resource interaction data and preset screening conditions, wherein the target resource interaction data includes resource interaction data of at least one of the cheating players and resource interaction data of a plurality of the players to be tested, wherein the preset screening conditions include an identifier of a known cheating player; According to the target resource interaction data, the resource interaction graph is constructed with the multiple game players as nodes and the resource interaction data flows between the multiple game players as edges.
3. The method according to claim 1, characterized in that: The step of screening out suspected cheating players from the plurality of players to be tested according to the group clustering result includes: Obtaining a degree of suspicion that each cluster is a cheating cluster, wherein the degree of suspicion is used to indicate a ratio of the number of players in the cluster marked as cheating to the total number of game players in the cluster; According to the suspicion degree of each cluster being a cheating cluster and a preset suspicion threshold, suspected cheating players are screened out from the multiple players to be tested.
4. The method according to claim 3, characterized in that The step of obtaining a representation vector of each node in the resource interaction graph according to a preset graph node embedding algorithm includes: A random walk-based graph embedding algorithm is used to extract the contextual relationship of each node in the resource interaction graph, where the contextual relationship of each node is used to indicate the upstream node and the downstream node of each node in the random walk path; According to the contextual relationship of each node in the resource interaction graph, a first representation vector of each node in the resource interaction graph is obtained.
5. The method according to claim 4, characterized in that The step of obtaining a representation vector of each node in the resource interaction graph according to a preset graph node embedding algorithm includes: According to the deep autoencoder algorithm, a second representation vector of each node in the resource interaction graph is obtained.
6. The method according to claim 5, characterized in that The method of generating a group clustering result of similar nodes by using a preset clustering algorithm according to the representation vector of each node includes: Generate a first group clustering result and / or a second group clustering result of similar nodes using a preset clustering algorithm according to the first characterization vector and / or the second characterization vector of each node; Accordingly, screening out suspected cheating players from the plurality of players to be tested according to the group clustering result includes: According to the first group clustering result and / or the second group clustering result, suspected cheating players are screened out from the multiple players to be tested.
7. The method according to claim 6, characterized in that The step of screening out suspected cheating players from the plurality of players to be tested according to the first group clustering result and / or the second group clustering result includes: Obtaining a first suspicion degree that each of the clusters in the first group clustering result is a plug-in cluster and / or a second suspicion degree that each of the clusters in the second group clustering result is a plug-in cluster; According to the first suspicion degree and / or the second suspicion degree and a preset suspicion threshold, suspected cheating players are screened out from the multiple players to be tested.
8. The method according to claim 6, characterized in that After the preset clustering algorithm is used to generate the first group clustering result and / or the second group clustering result of similar nodes, the method further includes: The first group clustering result and / or the second group clustering result are displayed according to a preset display mode.
9. A game cheat detection device, characterized in that: include: A first acquisition module, a second acquisition module, a generation module, and a screening module; The first acquisition module is used to acquire a resource interaction graph based on the target game, wherein the resource interaction graph includes a graph constructed with multiple game players as nodes and resource interaction data flows between the multiple game players as edges, wherein the multiple game players include at least one cheat player and multiple players to be tested; The second acquisition module is used to acquire the representation vector of each node in the resource interaction graph according to a preset graph node embedding algorithm; The generating module is used to generate a group clustering result of similar nodes by using a preset clustering algorithm according to the representation vector of each node; The screening module is used to screen out suspected cheating players from the plurality of players to be tested according to the group clustering result; The generating module is specifically used to calculate the similarity between the nodes using a preset clustering algorithm according to the first characterization vector and / or the second characterization vector of each node, the first characterization vector of each node is used to indicate the edge relationship between adjacent nodes, and the second characterization vector of each node is used to indicate the edge relationship between the node and surrounding nodes; According to the similarity between the nodes, the plurality of nodes are divided into a plurality of clusters, each of the clusters including at least a preset number of nodes.
10. An electronic device, characterized in that: include: A processor, a storage medium and a bus, wherein the storage medium stores machine-readable instructions executable by the processor. When the electronic device is running, the processor communicates with the storage medium via the bus, and the processor executes the machine-readable instructions to perform the steps of the game plug-in detection method as described in any one of claims 1-8.
11. A storage medium, characterized in that: The storage medium stores a computer program, and when the computer program is executed by the processor, the steps of the game plug-in detection method as claimed in any one of claims 1 to 8 are executed.
