Abnormal role identification methods, devices, equipment, storage media, and program products

Abstract

This application provides a method, apparatus, device, storage medium, and program product for identifying abnormal characters. The method includes: extracting target game feature values ​​corresponding to at least two target characters; wherein the target characters are characters from different factions participating in a target game match; inputting the target game feature values ​​into a pre-trained logistic regression prediction model to determine outliers between the target characters; the logistic regression prediction model is trained using the pre-acquired abnormal game feature values ​​of the abnormal characters; determining whether the outlier is greater than a preset outlier threshold; and, in response to determining that the outlier is greater than the preset outlier threshold, identifying the target character as an abnormal character. By using a logistic regression prediction model to determine outliers and thus whether the corresponding target character is an abnormal character, the accuracy of abnormal character identification is improved, further enhancing the user's gaming experience and improving the balance of the game ecosystem.

Description

Technical Field

[0001] This application relates to the field of computer technology, and in particular to an abnormal role recognition method, apparatus, device, storage medium, and program product. Background Technology

[0002] In related technologies, instances of illegal teaming among players exist in game scenarios. For example, in solo mode, players who were originally competitors may team up, leading to a poor gaming experience for other non-cooperative players. However, current technologies typically identify these illegal teaming players through videos recorded by other players. This method suffers from low accuracy and efficiency, resulting in a poor player experience and an imbalance in the game ecosystem. Summary of the Invention

[0003] In view of this, this application proposes an abnormal role identification method, apparatus, device, storage medium, and program product.

[0004] In a first aspect, this application provides an abnormal role identification method, the method comprising:

[0005] Extract target game feature values ​​corresponding to at least two target characters; wherein, the target characters are characters from different factions who participate in the target game match together;

[0006] The target game feature values ​​are input into a pre-trained logistic regression prediction model to determine outliers among the target characters; the logistic regression prediction model is trained using the pre-acquired abnormal game feature values ​​of the abnormal characters.

[0007] Determine whether the outlier is greater than a preset outlier threshold;

[0008] In response to determining that the outlier is greater than a preset outlier threshold, the target role is identified as an outlier role.

[0009] In a second aspect, this application provides an abnormal role recognition device, the device comprising:

[0010] The extraction module is configured to extract target game feature values ​​corresponding to at least two target characters; wherein, the target characters are characters from different factions who participate in the target game match together;

[0011] The prediction module is configured to input the target game feature values ​​into a pre-trained logistic regression prediction model to determine outliers among the target characters; the logistic regression prediction model is trained using the pre-acquired abnormal game feature values ​​of the abnormal characters.

[0012] The determination module is configured to determine whether the outlier is greater than a preset outlier threshold.

[0013] The identification module is configured to identify the target role as an abnormal role in response to determining that the abnormal value is greater than a preset abnormal value threshold.

[0014] In a third aspect, this application provides an electronic device including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the program, implements the abnormal role recognition method as described in the first aspect.

[0015] In a fourth aspect, this application provides a non-transitory computer-readable storage medium storing computer instructions for causing a computer to perform the abnormal role recognition method as described in the first aspect.

[0016] Fifthly, this application provides a computer program product, including computer program instructions that, when executed on a computer, cause the computer to perform the abnormal role recognition method as described in the first aspect.

[0017] As can be seen from the above, the abnormal character identification method, apparatus, device, storage medium, and program product provided in this application can extract target game feature values ​​corresponding to at least two target characters. These target characters are characters from different factions participating in the same target game match; that is, the target characters do not have a team game relationship and are independent of each other. Furthermore, the target game feature values ​​can be input into a pre-trained logistic regression prediction model, which then outputs anomalies between target users. The logistic regression prediction model can be trained using the pre-acquired abnormal game feature values ​​of the abnormal characters. Further, it can be determined whether the anomaly value is greater than a preset anomaly value threshold. If the anomaly value is determined to be greater than the preset anomaly value threshold, then at least two identified target characters are determined to be abnormal characters. By extracting the target game feature values ​​of characters from different factions participating in the same target game match, autonomous detection of abnormal characters can be achieved, improving the efficiency of abnormal character identification. Furthermore, by using the logistic regression prediction model to determine the anomalies, the accuracy of abnormal character identification is improved, further enhancing the user's gaming experience and improving the balance of the game ecosystem. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in this application or related technologies, the drawings used in the description of the embodiments or related technologies 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 these drawings without creative effort.

[0019] Figure 1 An exemplary flowchart of an abnormal role recognition method provided in an embodiment of this application is shown.

[0020] Figure 2 An exemplary structural diagram of an abnormal role recognition device provided in an embodiment of this application is shown.

[0021] Figure 3 This illustration shows an exemplary structural diagram of an electronic device provided in an embodiment of this application. Detailed Implementation

[0022] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with specific embodiments and the accompanying drawings.

[0023] It should be noted that, unless otherwise defined, the technical or scientific terms used in the embodiments of this application should have the ordinary meaning understood by one of ordinary skill in the art to which this application pertains. The terms "first," "second," and similar terms used in the embodiments of this application do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the elements or objects listed after the word and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are only used to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0024] As described in the background section, there are instances of illegal team formation between players in the game. For example, in solo mode, players who were previously in competition may have a poor gaming experience for other non-cooperative players because some players are playing solo mode in a cooperative manner.

[0025] The inventors' research revealed that in related technologies, such as those used in gaming scenarios, these illegally teamed players might be two acquaintances who deliberately choose a time with fewer players to be matched in the same game, or they might communicate during the game using coded language commonly used in the illegal teaming community. Then, these two or more players, acting as a team, would confuse and defeat other players, ultimately achieving a higher ranking and better in-game rewards. This situation is extremely detrimental to the gaming environment. However, related technologies typically rely on videos recorded by other players to identify these illegal teaming players, resulting in low accuracy and efficiency in identification. This leads to a poor player experience and an imbalance in the game ecosystem.

[0026] Therefore, this application provides an abnormal character identification method, apparatus, device, storage medium, and program product, which can extract target game feature values ​​corresponding to at least two target characters. The target characters are characters from different factions participating in the same target game match; that is, the target characters do not have a team game relationship and are independent of each other. Further, the target game feature values ​​can be input into a pre-trained logistic regression prediction model, which then outputs anomalies between target users. The logistic regression prediction model can be trained using the pre-acquired abnormal game feature values ​​of the abnormal characters. Further still, it can be determined whether the anomaly value is greater than a preset anomaly value threshold. If the anomaly value is determined to be greater than the preset anomaly value threshold, then at least two identified target characters are determined to be abnormal characters. By extracting the target game feature values ​​of characters from different factions participating in the same target game match, autonomous detection of abnormal characters can be achieved, improving the efficiency of abnormal character identification. Furthermore, by using the logistic regression prediction model to determine the anomalies, the accuracy of abnormal character identification is improved, further enhancing the user's gaming experience and improving the balance of the game ecosystem.

[0027] The abnormal role recognition method provided in this application will be specifically described below through specific embodiments.

[0028] Figure 1 An exemplary flowchart of an abnormal role recognition method provided in an embodiment of this application is shown.

[0029] refer to Figure 1 The abnormal role recognition method provided in this application specifically includes the following steps:

[0030] S102: Extract target game feature values ​​corresponding to at least two target characters; wherein, the target characters are characters from different factions who participate in the target game match together.

[0031] S104: Input the target game feature values ​​into a pre-trained logistic regression prediction model to determine outliers among the target characters; the logistic regression prediction model is trained using the pre-acquired abnormal game feature values ​​of the abnormal characters.

[0032] S106: Determine whether the outlier is greater than a preset outlier threshold.

[0033] S108: In response to determining that the outlier is greater than a preset outlier threshold, the target role is determined to be an outlier role.

[0034] In some embodiments, the identification of anomalous characters can be targeted at solo mode in a game match. For example, in a game match with only one winner, there are 100 virtual characters, at least some of which are controlled by real players via terminals. Each virtual character is independent, operating independently, with each pair of characters forming opposing factions. Therefore, in this game mode, according to the original game logic, each character should be an enemy. However, if at least two characters in the same game match have an illegal team relationship—that is, at least two characters cooperate against other characters in the same game match—then the characters with the illegal team relationship are identified as anomalous characters. Therefore, since each character generates corresponding game logs when participating in a game match, to identify these anomalous characters within the game match, characters from different factions participating in the target game match can be identified as target characters, and their corresponding game logs can be considered valid game logs.

[0035] It should be noted that even if there is at least one virtual character (e.g., virtual character X) controlled by a virtual player created by the system among the 100 virtual characters, since the virtual player is performing game behaviors specified by the system, it can be determined that this type of virtual character X controlled by the virtual player is in accordance with the game logic. That is, this type of virtual character X will not cause the aforementioned illegal teaming situation. In other words, it is guaranteed that the identified abnormal characters are virtual characters controlled by real users.

[0036] Abnormal game characteristics can include one or more of the following: average distance abnormal characteristics, mutual damage abnormal characteristics, or assist damage abnormal characteristics. Since abnormal users are usually characters involved in illegal team formations, these characters may be close together, but the damage they inflict on each other is low. Furthermore, when character A defeats character B, character C, who is also an opponent of character B, inflicts a large amount of damage on character B during the defeat. Therefore, this can be used as a basis for determining abnormal game characteristics.

[0037] Specifically, the average distance anomaly characteristic value can be determined by extracting scene distance data between multiple abnormal characters from the game logs of the target game at preset time intervals, such as one second, and averaging these obtained scene distance data. It should be noted that this scene distance data can be determined by the relative distance between the centroids of two abnormal characters in the game scene. For the mutual damage anomaly characteristic value, the damage values ​​inflicted between at least two abnormal characters within a preset number of matches, such as ten matches, can be extracted from the game logs. For the assist damage characteristic value, the damage values ​​inflicted by at least two abnormal characters on the same other character within a preset number of matches can be extracted from the game logs. This is used as the percentage of assist damage caused by each abnormal character defeating the other, i.e., the assist damage characteristic value. Other characters can be any characters participating in the target game other than the target character.

[0038] It should be noted that the average distance anomaly characteristic value can be determined by setting an average distance anomaly characteristic value threshold. For example, if the average distance anomaly characteristic value threshold is set to 10 meters, then if two target characters are within 10 meters of each other and the relative distance remains within 10 meters for a preset time (e.g., 2 minutes), and neither target character has defeated the other, then this distance data of 10 meters can be used as the average distance anomaly characteristic value between these two target characters who are considered to be exhibiting abnormal behavior. Alternatively, the average distance between characters identified as exhibiting abnormal behavior in the same game match can be extracted from historical game logs and used as the average distance anomaly characteristic value.

[0039] In some embodiments, to ensure that the obtained abnormal game feature values ​​better match the prediction model algorithm, they can be normalized. For example, when there are three types of abnormal game feature values, each type can be normalized separately to determine the target abnormal game feature value. Specifically, for each type of abnormal game feature value, the maximum and minimum abnormal game feature values ​​can be determined. For example, taking the average distance abnormal feature value as an example, the maximum and minimum average distance abnormal feature values ​​in the obtained sample data can be determined.

[0040] Furthermore, the abnormal game feature value corresponding to any abnormal character, i.e., the average distance abnormal feature value, can be determined. The first difference between the abnormal game feature value and the minimum abnormal game feature value of this abnormal character, and the second difference between the maximum and minimum abnormal game feature values, can also be determined. In other words, the first difference between the average distance abnormal feature value and the minimum average distance abnormal feature value of this abnormal character, and the second difference between the maximum and minimum average distance abnormal feature values, can be determined, and the target abnormal game feature value can be determined based on the ratio of the first difference and the second difference. For example, the target abnormal game feature value can be expressed as...

[0041]

[0042] Where X represents the abnormal game feature value, min represents the minimum abnormal game feature value, and max represents the maximum abnormal game feature value. new This represents the target abnormal game feature value.

[0043] It should be noted that, regardless of whether the abnormal game feature is one or more of the following: average distance abnormal feature, mutual damage abnormal feature, or assist damage abnormal feature, each abnormal game feature can be normalized using the above formula to determine the target abnormal game feature.

[0044] In some embodiments, a corresponding weight parameter can be set for each target abnormal game feature value. This weight parameter can be used to indicate the importance of the corresponding target abnormal game feature value to the identification result; that is, the larger the weight parameter, the greater the impact of the target abnormal game feature value on the identification result. The weight parameter can be determined based on existing models or manually set according to actual application scenarios. The specific method of determining the weight parameter is not specifically limited. A logistic regression prediction model can be constructed based on the target abnormal game feature values ​​and their corresponding weight parameters. For example, when the target abnormal game feature values ​​include average distance abnormal feature values, mutual damage abnormal feature values, and assist damage abnormal feature values, the logistic regression prediction model can be expressed as follows:

[0045] Y = a*X1 + β*X2 + γ*X3

[0046] Where Y represents the logistic regression prediction model, α represents the first weight parameter, β represents the second weight parameter, γ represents the third weight parameter, X1 represents the normalized average distance anomaly feature value between anomaly characters, X2 represents the normalized mutual damage anomaly feature value between anomaly characters, and X3 represents the normalized assist damage anomaly feature value between anomaly characters.

[0047] It should be noted that more abnormal game feature values ​​and corresponding weight parameters can be added to the logistic regression prediction model formula above, depending on the actual application scenario. Here, α can be 0.8, β can be 1.2, and γ can be 1.5.

[0048] Specifically, the training process of the logistic regression prediction model for predicting anomalous users is as follows: Abnormal game feature values ​​corresponding to characters already identified as anomalous can be extracted from the game logs of the target game. For example, a group of users confirmed to be in illegal team play can be identified, and the game logs generated by the characters controlled by these users in the target game (e.g., game logs in solo mode) can be extracted. These characters are considered as identified anomalous characters and used as positive samples for the prediction model. Normal game feature values ​​corresponding to normal characters confirmed to have no anomalous behavior can be obtained from the game logs and used as negative samples for the logistic regression prediction model. It can be understood that positive samples represent anomalous characters, and negative samples represent normal characters. After constructing the logistic regression prediction model, prediction tests can be performed on the positive samples in the sample data, i.e., the abnormal game feature values. The known abnormal game feature values ​​are input into the logistic regression prediction model to obtain the training prediction results. Furthermore, the prediction precision and prediction recall of the logistic regression prediction model can be determined based on the training prediction results. For example, referring to Table 1, the number of data points that are actually positive samples and whose training prediction results are also positive is A, for example, A = 300; the number of data points that are actually positive samples but whose training prediction results are negative is C, for example, C = 200; the number of data points that are actually negative samples but whose training prediction results are positive is B, for example, B = 100; and the number of data points that are actually negative samples and whose training prediction results are also negative is D, for example, D = 400. Then, the prediction precision can be expressed as P = A / (A+B) = 300 / (300+100) = 0.75, and the prediction recall can be expressed as R = A / (A+C) = 300 / (300+200) = 0.6. For example, when the prediction precision reaches 0.7 or higher, and the prediction recall reaches 0.5 or higher, then a well-trained logistic regression prediction model is obtained.

[0049] Table 1. Confusion matrix for positive and negative samples.

[0050] Actually a positive sample Actually a negative sample The model predicts positive samples. A B The model predicts negative samples. C D

[0051] In some embodiments, when using a pre-established logistic regression prediction model to predict whether a target character is an anomalous character, target game feature values ​​corresponding to the target character can be obtained. For example, the average distance anomalous feature value can be determined by extracting the relative distances between at least two target characters from game logs at preset time intervals, such as one second, and averaging these obtained relative distances. For mutual damage anomalous feature values, the damage values ​​caused to each other by at least two target characters within a preset number of matches, such as ten matches, can be extracted from game logs. For assist damage feature values, the damage values ​​caused to the same other character by virtual characters corresponding to at least two target characters within a preset number of matches can be extracted from game logs, serving as the percentage of assist damage caused by each target character defeating the other, i.e., the assist damage feature value.

[0052] Furthermore, a logistic regression prediction model is used to predict the input target game feature values ​​to identify outliers. By determining whether the outliers exceed a preset outlier threshold, it is determined whether the target character is an anomalous character. It should be noted that the preset outlier threshold can be determined using the F1-score (FI score).

[0053] In some embodiments, using a trained logistic regression prediction model, the prediction precision of the model can be expressed as P = A / (A+B) = 300 / (300+100) = 0.75, and the prediction recall can be expressed as R = A / (A+C) = 300 / (300+200) = 0.6.

[0054] Furthermore, a first threshold intermediate parameter can be determined by combining the product of prediction precision and prediction recall with a preset multiple, and a second threshold intermediate parameter can be determined by summing prediction precision and prediction recall. Finally, a preset outlier threshold can be determined based on the first and second threshold intermediate parameters. Specifically, the preset outlier threshold can be expressed as follows:

[0055]

[0056] Where P represents prediction precision and R represents prediction recall. Then, by calculating each threshold (0 to 1, divided by 0.01), the case with the largest F1-Score is found, and the threshold at this time is retained as the standard. For example, when P = 0.75 and R = 0.6, the determined F1-score is 0.67, and the preset outlier threshold can be set to 0.6.

[0057] In some embodiments, after determining the preset outlier threshold and the outlier value, if the outlier value is less than or equal to the preset outlier threshold, it can be determined that the target character does not have abnormal behavior, that is, the target character is not an abnormal character, and the target character can still be assigned to the matching pool of normal characters to play the target game together with normal characters. However, if the outlier value is greater than the preset outlier threshold, it can be determined that the target character has abnormal behavior, that is, the target character is an abnormal character, and the abnormal game account corresponding to the abnormal character can be assigned to a restricted matching pool. The restricted matching pool only includes the abnormal game account corresponding to the abnormal character, thereby isolating the abnormal character from other normal characters and ensuring that the abnormal character will not be matched with other normal characters, thus avoiding the abnormal character affecting the game experience of other normal characters.

[0058] It should be noted that the identification of abnormal characters in cases of illegal teaming can also be applied to multiplayer team game modes. In this case, the target character is still a character from a different faction who is participating in the same game. However, compared to solo mode, since there are multiple characters in the same team or faction in multiplayer team mode, it is more difficult to determine which character is illegally teaming up with other factions or teams. Therefore, abnormal characters in multiplayer team mode may be identified as an entire group of characters. For example, if there are a maximum of 4 game characters in each faction or team, then when at least one character in a faction or team is identified as an abnormal character, all characters in the faction or team to which the abnormal character belongs will be identified as abnormal characters.

[0059] As can be seen from the above, the abnormal character identification method, apparatus, device, storage medium, and program product provided in this application can extract target game feature values ​​corresponding to at least two target characters. These target characters are characters from different factions participating in the same target game match; that is, the target characters do not have a team game relationship and are independent of each other. Furthermore, the target game feature values ​​can be input into a pre-trained logistic regression prediction model, which then outputs anomalies between target users. The logistic regression prediction model can be trained using the pre-acquired abnormal game feature values ​​of the abnormal characters. Further, it can be determined whether the anomaly value is greater than a preset anomaly value threshold. If the anomaly value is determined to be greater than the preset anomaly value threshold, then at least two identified target characters are determined to be abnormal characters. By extracting the target game feature values ​​of characters from different factions participating in the same target game match, autonomous detection of abnormal characters can be achieved, improving the efficiency of abnormal character identification. Furthermore, by using the logistic regression prediction model to determine the anomalies, the accuracy of abnormal character identification is improved, further enhancing the user's gaming experience and improving the balance of the game ecosystem.

[0060] It should be noted that the method in this embodiment can be executed by a single device, such as a computer or server. The method can also be applied in a distributed scenario, where multiple devices cooperate to complete the task. In such a distributed scenario, one of these devices may execute only one or more steps of the method in this embodiment, and the multiple devices will interact with each other to complete the method described.

[0061] It should be noted that the above description describes some embodiments of this application. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in a different order than that shown in the above embodiments and still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0062] Figure 2 An exemplary structural diagram of an abnormal role recognition device provided in an embodiment of this application is shown.

[0063] Based on the same inventive concept, corresponding to any of the above embodiments, this application also provides an abnormal role recognition device.

[0064] refer to Figure 2 The abnormal role recognition device includes: an extraction module, a prediction module, a determination module, and a recognition module; wherein,

[0065] The extraction module is configured to extract target game feature values ​​corresponding to at least two target characters; wherein, the target characters are characters from different factions who participate in the target game match together;

[0066] The prediction module is configured to input the target game feature values ​​into a pre-trained logistic regression prediction model to determine outliers among the target characters; the logistic regression prediction model is trained using the pre-acquired abnormal game feature values ​​of the abnormal characters.

[0067] The determination module is configured to determine whether the outlier is greater than a preset outlier threshold.

[0068] The identification module is configured to identify the target role as an abnormal role in response to determining that the abnormal value is greater than a preset abnormal value threshold.

[0069] In one possible implementation, the target game feature value includes: the average scene distance feature value between the target characters;

[0070] The extraction module is further configured to:

[0071] Based on the game log of the target game, scene distance data between multiple target characters is extracted at multiple different times in the target game to determine multiple scene distance data;

[0072] The average scene distance of the multiple scene distance data is determined to obtain the average scene distance feature value between the target characters.

[0073] In one possible implementation, the target game feature value further includes: a mutual damage feature value used to indicate the damage caused to each other by the target characters;

[0074] The extraction module is further configured to:

[0075] Based on the game logs of the target game matches, extract the damage values ​​inflicted between at least two target characters within a preset number of matches to determine the mutual damage characteristic values ​​between the target characters.

[0076] In one possible implementation, the target game feature value further includes: an assist damage feature value indicating the damage dealt by the target character to the same other character participating in the target game; the other character is a character other than the target character participating in the target game.

[0077] The extraction module is further configured to:

[0078] Based on the game logs of the target game, extract the damage values ​​caused by at least two target characters to the same other character participating in the target game within a preset number of matches to determine the assist damage characteristic value between the target characters.

[0079] In one possible implementation, the device further includes: a construction module;

[0080] The building module is configured as follows:

[0081] Extract abnormal game feature values ​​corresponding to abnormal characters from the game logs of the target game; wherein, the abnormal game feature values ​​include at least one of the following: the average distance abnormal feature value between the abnormal characters, the mutual damage abnormal feature value used to indicate that the abnormal characters cause damage to each other, or the assist damage abnormal feature value used to indicate that the abnormal characters cause damage to the same target character.

[0082] The abnormal game feature values ​​are normalized respectively to determine the target abnormal game feature values;

[0083] The weight parameters corresponding to the target abnormal game feature values ​​are set respectively, and the logistic regression prediction model is constructed based on the target abnormal game feature values ​​and the weight parameters.

[0084] In one possible implementation, the building module is further configured as follows:

[0085] Determine the maximum and minimum abnormal game feature values ​​among the abnormal game feature values.

[0086] Determine the first difference between the abnormal game feature value corresponding to any abnormal character and the minimum abnormal game feature value, and the second difference between the maximum abnormal game feature value and the minimum abnormal game feature value, and determine the target abnormal game feature value based on the ratio of the first difference and the second difference.

[0087] In one possible implementation, the logistic regression prediction model includes: weight parameters corresponding to the target game feature values; the weight parameters are used to characterize the correlation between the target game feature values ​​and the outliers;

[0088] The determining module is further configured to:

[0089] Each target game feature value is determined by multiplying it with its corresponding weight parameter, and outliers among the target characters are determined by summing each product.

[0090] In one possible implementation, the device further includes: a second determining module;

[0091] The second determining module is configured as follows:

[0092] The abnormal characters corresponding to the abnormal game feature values ​​are predicted based on the logistic regression prediction model to output the training prediction results.

[0093] The prediction precision and prediction recall of the logistic regression prediction model are determined based on the training prediction results.

[0094] The intermediate parameters of the first threshold and the intermediate parameters of the second threshold are determined based on the prediction precision and prediction recall, respectively.

[0095] The preset outlier threshold is determined based on the first threshold intermediate parameter and the second threshold intermediate parameter.

[0096] For ease of description, the above devices are described in terms of function, divided into various modules. Of course, in implementing this application, the functions of each module can be implemented in one or more software and / or hardware.

[0097] The apparatus described above is used to implement the corresponding abnormal role recognition method in any of the foregoing embodiments, and has the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0098] Figure 3 This illustration shows an exemplary structural diagram of an electronic device provided in an embodiment of this application.

[0099] Based on the same inventive concept, corresponding to the methods of any of the above embodiments, this application also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the abnormal role recognition method described in any of the above embodiments. Figure 3 This embodiment illustrates a more specific hardware structure of an electronic device, which may include a processor 310, a memory 320, an input / output interface 330, a communication interface 340, and a bus 350. The processor 310, memory 320, input / output interface 330, and communication interface 340 are interconnected internally via the bus 350.

[0100] The processor 310 can be implemented using a general-purpose CPU (Central Processing Unit), microprocessor, application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this specification.

[0101] The memory 320 can be implemented in the form of ROM (Read Only Memory), RAM (Random Access Memory), static storage device, dynamic storage device, etc. The memory 320 can store the operating system and other applications. When the technical solutions provided in the embodiments of this specification are implemented by software or firmware, the relevant program code is stored in the memory 320 and is called and executed by the processor 310.

[0102] Input / output interface 330 is used to connect input / output modules to realize information input and output. Input / output modules can be configured as components in the device (not shown in the figure) or externally connected to the device to provide corresponding functions. Input devices may include keyboards, mice, touch screens, microphones, various sensors, etc., and output devices may include displays, speakers, vibrators, indicator lights, etc.

[0103] The communication interface 340 is used to connect a communication module (not shown in the figure) to enable communication between this device and other devices. The communication module can communicate via wired means (such as USB, Ethernet cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.).

[0104] Bus 350 includes a pathway for transmitting information between various components of the device (e.g., processor 310, memory 320, input / output interface 330, and communication interface 340).

[0105] It should be noted that although the above-described device only shows the processor 310, memory 320, input / output interface 330, communication interface 340, and bus 350, in specific implementations, the device may also include other components necessary for normal operation. Furthermore, those skilled in the art will understand that the above-described device may only include the components necessary for implementing the embodiments of this specification, and not necessarily all the components shown in the figures.

[0106] The electronic devices described above are used to implement the corresponding abnormal role recognition methods in any of the foregoing embodiments, and have the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0107] Based on the same inventive concept, corresponding to the methods of any of the above embodiments, this application also provides a non-transitory computer-readable storage medium that stores computer instructions for causing the computer to execute the abnormal role recognition method as described in any of the above embodiments.

[0108] The computer-readable medium of this embodiment includes permanent and non-permanent, removable and non-removable media, and information storage can be implemented by any method or technology. Information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic magnetic disk storage or other magnetic storage devices, or any other non-transfer medium that can be used to store information accessible by a computing device.

[0109] The computer instructions stored in the storage medium of the above embodiments are used to cause the computer to execute the abnormal role recognition method as described in any of the above embodiments, and have the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0110] Based on the same inventive concept, corresponding to the abnormal role recognition method described in any of the above embodiments, this disclosure also provides a computer program product, which includes computer program instructions. In some embodiments, the computer program instructions can be executed by one or more processors of a computer to cause the computer and / or the processor to perform the abnormal role recognition method. Corresponding to the execution entity for each step in each embodiment of the abnormal role recognition method, the processor executing the corresponding step can belong to the corresponding execution entity.

[0111] The computer program products of the above embodiments are used to cause the computer and / or the processor to execute the abnormal role recognition method as described in any of the above embodiments, and have the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0112] Those skilled in the art should understand that the discussion of any of the above embodiments is merely exemplary and is not intended to imply that the scope of this application (including the claims) is limited to these examples; within the framework of this application, the technical features of the above embodiments or different embodiments can also be combined, the steps can be implemented in any order, and there are many other variations of different aspects of the embodiments of this application as described above, which are not provided in the details for the sake of brevity.

[0113] Additionally, to simplify the description and discussion, and to avoid obscuring the embodiments of this application, the well-known power / ground connections to integrated circuit (IC) chips and other components may or may not be shown in the provided drawings. Furthermore, the apparatus may be shown in block diagram form to avoid obscuring the embodiments of this application, and this also takes into account the fact that the details of the implementation of these block diagram apparatuses are highly dependent on the platform on which the embodiments of this application will be implemented (i.e., these details should be fully understood by those skilled in the art). While specific details (e.g., circuits) have been set forth to describe exemplary embodiments of this application, it will be apparent to those skilled in the art that the embodiments of this application can be implemented without these specific details or with variations thereof. Therefore, these descriptions should be considered illustrative rather than restrictive.

[0114] Although this application has been described in conjunction with specific embodiments thereof, many substitutions, modifications, and variations of these embodiments will be apparent to those skilled in the art from the foregoing description. For example, other memory architectures (e.g., dynamic RAM (DRAM)) may be used with the embodiments discussed.

[0115] The embodiments of this application are intended to cover all such substitutions, modifications, and variations that fall within the broad scope of the appended claims. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the embodiments of this application should be included within the protection scope of this application.

Claims

1. An abnormal role recognition method, characterized in that, The method includes: Extract target game feature values ​​corresponding to at least two target characters from the game logs of the target game; wherein, the target characters are characters from different factions participating in the target game; the target game feature values ​​include: the average scene distance feature value between the target characters; the mutual damage feature value indicating the damage caused to each other by the target characters; and the assist damage feature value indicating the damage caused by the target character to the same other character participating in the target game; the other character is any character other than the target character participating in the target game. The extraction of target game feature values ​​corresponding to at least two target characters includes: Based on the game log of the target game, scene distance data between multiple target characters is extracted at multiple different times in the target game to determine multiple scene distance data; Determine the average scene distance of the multiple scene distance data to obtain the average scene distance feature value between the target characters; The target game feature values ​​are input into a pre-trained logistic regression prediction model to determine outliers among the target characters; the logistic regression prediction model is trained using the pre-acquired abnormal game feature values ​​of the abnormal characters. Determine whether the outlier is greater than a preset outlier threshold; In response to determining that the outlier is greater than a preset outlier threshold, the target character is identified as an outlier, and the outlier game account corresponding to the outlier is assigned to a restricted matchmaking pool.

2. The method according to claim 1, characterized in that, The extraction of target game feature values ​​corresponding to at least two target characters includes: Based on the game logs of the target game matches, extract the damage values ​​inflicted between at least two target characters within a preset number of matches to determine the mutual damage characteristic values ​​between the target characters.

3. The method according to claim 2, characterized in that, The extraction of target game feature values ​​corresponding to at least two target characters includes: Based on the game logs of the target game, extract the damage values ​​caused by at least two target characters to the same other character participating in the target game within a preset number of matches to determine the assist damage characteristic value between the target characters.

4. The method according to claim 1, characterized in that, Before inputting the target game feature values ​​into a pre-trained logistic regression prediction model to determine outliers among the target characters, the method further includes: Extract abnormal game feature values ​​corresponding to abnormal characters from the game logs of the target game; wherein, the abnormal game feature values ​​include at least one of the following: the average distance abnormal feature value between the abnormal characters, the mutual damage abnormal feature value used to indicate that the abnormal characters cause damage to each other, or the assist damage abnormal feature value used to indicate that the abnormal characters cause damage to the same target character. The abnormal game feature values ​​are normalized respectively to determine the target abnormal game feature values; The weight parameters corresponding to the target abnormal game feature values ​​are set respectively, and the logistic regression prediction model is constructed based on the target abnormal game feature values ​​and the weight parameters.

5. The method according to claim 4, characterized in that, The normalization process for the abnormal game feature values ​​to determine the target abnormal game feature values ​​includes: Determine the maximum and minimum abnormal game feature values ​​among the abnormal game feature values. Determine the first difference between the abnormal game feature value corresponding to any abnormal character and the minimum abnormal game feature value, and the second difference between the maximum abnormal game feature value and the minimum abnormal game feature value, and determine the target abnormal game feature value based on the ratio of the first difference and the second difference.

6. The method according to claim 1, characterized in that, The logistic regression prediction model includes: weight parameters corresponding to the target game feature values; the weight parameters are used to characterize the correlation between the target game feature values ​​and the outliers; The step of inputting the target game feature values ​​into a pre-trained logistic regression prediction model to determine outliers among the target characters includes: Each target game feature value is determined by multiplying it with its corresponding weight parameter, and outliers among the target characters are determined by summing each product.

7. The method according to claim 4, characterized in that, After constructing the logistic regression prediction model based on the target abnormal game feature values ​​and the weight parameters, the method further includes: The abnormal characters corresponding to the abnormal game feature values ​​are predicted based on the logistic regression prediction model to output the training prediction results. The prediction precision and prediction recall of the logistic regression prediction model are determined based on the training prediction results. The intermediate parameters of the first threshold and the intermediate parameters of the second threshold are determined based on the prediction precision and prediction recall, respectively. The preset outlier threshold is determined based on the first threshold intermediate parameter and the second threshold intermediate parameter.

8. An abnormal role recognition device, characterized in that, The device includes: The extraction module is configured to extract target game feature values ​​corresponding to at least two target characters based on the game logs of the target game match; wherein, the target characters are characters from different factions participating in the target game match; the target game feature values ​​include: an average scene distance feature value between the target characters; a mutual damage feature value indicating damage caused to each other by the target characters; and an assist damage feature value indicating the damage caused by the target character to the same other character participating in the target game match; the other character is a character other than the target character participating in the target game match. The extraction of target game feature values ​​corresponding to at least two target characters includes: Based on the game log of the target game, scene distance data between multiple target characters is extracted at multiple different times in the target game to determine multiple scene distance data; Determine the average scene distance of the multiple scene distance data to obtain the average scene distance feature value between the target characters; The prediction module is configured to input the target game feature values ​​into a pre-trained logistic regression prediction model to determine outliers among the target characters; the logistic regression prediction model is trained using the pre-acquired abnormal game feature values ​​of the abnormal characters. The determination module is configured to determine whether the outlier is greater than a preset outlier threshold. The identification module is configured to identify the target character as an abnormal character and assign the abnormal game account corresponding to the abnormal character to the restricted match pool in response to determining that the abnormal value is greater than a preset abnormal value threshold.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the method as described in any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium storing computer instructions, characterized in that, The computer instructions are used to cause the computer to implement the method according to any one of claims 1 to 7.

11. A computer program product comprising computer program instructions that, when executed on a computer, cause the computer to perform the method as described in any one of claims 1 to 7.

Images