Account association and identification method, device and equipment, and storage medium

Abstract

The application provides an account association and identification method, device, equipment and storage medium, through the provided account association method, different types of user feature data in each to-be-associated account are more comprehensively analyzed and processed, and the accuracy of associating different accounts of the same user is effectively improved; in addition, the application also provides an account identification method, under the condition that multiple game accounts of the same user in a target game are known, the problem of how to distinguish the master-slave relationship between different game accounts of the same user can be solved, which is beneficial to creating a more fair and just game atmosphere and promoting the benign development of games.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of information processing, in particular to a method and device for account association and identification, and a storage medium. BACKGROUND

[0002] A user can register multiple different application accounts in the same application program. The prior art mainly associates different application accounts under the same user name through user binding information (such as a mobile phone number or an email account submitted when registering an account) in different application accounts. However, since the user can submit different binding information (such as using different email accounts to register multiple accounts) when registering an account, the above-mentioned account association method based on unique user binding information can miss some information, thereby resulting in low accuracy of account association. SUMMARY

[0003] Therefore, the present application aims to provide a method and device for account association and identification, and a storage medium, so as to more comprehensively analyze and process different types of user feature data in each account to be associated, thereby effectively improving the accuracy of associating different accounts of the same user. In addition, the present application also provides a method for account identification, which can solve the problem of how to distinguish the master-slave relationship between different game accounts of the same user under the condition that multiple game accounts of the same user in a target game are known, thereby helping to create a more fair and just game atmosphere and promoting the healthy development of games.

[0004] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the following preferred embodiments are described in detail below, together with the accompanying drawings.

[0005] In a first aspect, the embodiments of the present application provide a method for account association, which comprises:

[0006] obtaining first user feature data of multiple accounts to be associated, and performing one-hot encoding on each first user feature data to obtain a first feature encoding result of each account to be associated;

[0007] obtaining second user feature data of each account to be associated, clustering all second user feature data according to the edit distance between different second user feature data of accounts to be associated to obtain a first clustering result, and performing one-hot encoding on each cluster in the first clustering result to obtain a second feature encoding result of each account to be associated;

[0008] clustering the multiple accounts to be associated according to the first feature encoding result and the second feature encoding result to obtain a second clustering result;

[0009] In each cluster in the second clustering result, the similarity between different to-be-associated accounts in the cluster is calculated, and a target associated account is determined from the plurality of to-be-associated accounts according to the similarity.

[0010] In a second aspect, an embodiment of the present application provides a method for identifying an account, the method comprising:

[0011] obtaining a plurality of game accounts associated with a same user in a target game;

[0012] obtaining user feature data and target game character feature data in each game account, and performing normalization processing on the obtained user feature data and character feature data to obtain account feature data of each game account;

[0013] inputting the account feature data into a pre-trained account score calculation model to output an account score calculation result for each game account, wherein the account score calculation result is used to represent the association degree between each game account and the user;

[0014] determining a main game account and a sub game account of the user in the target game according to the account score calculation result of each game account.

[0015] In a third aspect, an embodiment of the present application provides an account association device, the device comprising:

[0016] a first obtaining module configured to obtain first user feature data of a plurality of to-be-associated accounts, and perform one-hot encoding on each first user feature data to obtain a first feature encoding result of each to-be-associated account;

[0017] a second obtaining module configured to obtain second user feature data of each to-be-associated account, perform clustering on all second user feature data according to the edit distance between different to-be-associated account second user feature data to obtain a first clustering result, and perform one-hot encoding on each cluster in the first clustering result to obtain a second feature encoding result of each to-be-associated account;

[0018] a first processing module configured to perform clustering on the plurality of to-be-associated accounts according to the first feature encoding result and the second feature encoding result to obtain a second clustering result;

[0019] an association module configured to calculate the similarity between different to-be-associated accounts in each cluster in the second clustering result, and determine a target associated account from the plurality of to-be-associated accounts according to the similarity.

[0020] Fourthly, embodiments of this application provide an account recognition device, the account recognition device comprising:

[0021] The account acquisition module is used to acquire multiple game accounts associated with the same user in the target game;

[0022] The data processing module is used to obtain user feature data and target game character feature data from each game account, and to normalize the obtained user feature data and character feature data to obtain account feature data for each game account.

[0023] The score determination module is used to input the account feature data into a pre-trained account score calculation model and output the account score calculation result for each game account; wherein, the account score calculation result is used to characterize the degree of association between each game account and the user;

[0024] The account identification module is used to determine the user's main game account and sub-game account in the target game based on the account score calculation results of each game account.

[0025] Fifthly, embodiments of this application provide a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the account association method described above.

[0026] Sixthly, embodiments of this application provide a computer-readable storage medium storing a computer program, which, when executed by a processor, performs the steps of the account association method described above.

[0027] In a seventh aspect, embodiments of this application provide a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the account identification method described above.

[0028] Eighthly, embodiments of this application provide a computer-readable storage medium storing a computer program, which, when executed by a processor, performs the steps of the account identification method described above.

[0029] The technical solutions provided by the embodiments of this application may include the following beneficial effects:

[0030] This application provides an account association and identification method, apparatus, device, and storage medium. The provided account association method performs more comprehensive analysis and processing of different types of user characteristic data in each account to be associated, effectively improving the accuracy of associating different accounts of the same user. Furthermore, this application also provides an account identification method that, given multiple game accounts of the same user in a target game, can specifically address the problem of distinguishing the master-slave relationship between different game accounts of the same user, thus contributing to a fairer and more just gaming environment and promoting the healthy development of the game. Attached Figure Description

[0031] 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.

[0032] Figure 1 A flowchart illustrating an account association method provided in an embodiment of this application is shown;

[0033] Figure 2 The illustration shows a flowchart of a method for one-hot encoding of the IP address of an account to be associated, according to an embodiment of this application.

[0034] Figure 3 This illustration shows a flowchart of a method for sorting clusters contained in a first clustering result, according to an embodiment of this application.

[0035] Figure 4 This illustration shows a flowchart of a method for identifying the main game account of a user in a target associated account, according to an embodiment of this application.

[0036] Figure 5 This illustration shows a flowchart of a method for training an account score calculation model according to an embodiment of this application;

[0037] Figure 6 A flowchart illustrating an account identification method provided in an embodiment of this application is shown;

[0038] Figure 7 This paper shows a schematic diagram of the structure of an account association device provided in an embodiment of this application;

[0039] Figure 8 This paper shows a schematic diagram of the structure of an account recognition device provided in an embodiment of this application;

[0040] Figure 9This is a schematic diagram of the structure of an electronic device 900 provided in an embodiment of this application;

[0041] Figure 10 This is a schematic diagram of the structure of another electronic device 1000 provided in an embodiment of this application. Detailed Implementation

[0042] 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. It should be understood that the accompanying drawings in this application are for illustrative and descriptive purposes only and are not intended to limit the scope of protection of this application. Furthermore, it should be understood that the schematic drawings are not drawn to scale. The flowcharts used in this application illustrate operations implemented according to some embodiments of this application. It should be understood that the operations in the flowcharts may not be implemented in sequence, and steps without logical contextual relationships may be reversed or implemented simultaneously. In addition, those skilled in the art, guided by the content of this application, may add one or more other operations to the flowcharts, or remove one or more operations from the flowcharts.

[0043] Furthermore, the described embodiments are merely some, not all, of the embodiments of this application. The components of the embodiments of this application described and illustrated herein can typically be arranged and designed in various different configurations. 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.

[0044] It should be noted that the term "comprising" will be used in the embodiments of this application to indicate the presence of the features declared thereafter, but does not exclude the addition of other features.

[0045] Users may register multiple different application accounts within the same application. Existing technology primarily associates different application accounts under the same username using user-linked information (such as the phone number or email address submitted during account registration). However, since users may submit different linking information during account registration (e.g., registering multiple accounts using different email addresses), the aforementioned method of account association based on unique user-linked information may miss some entries, resulting in low accuracy in account association.

[0046] Based on this, embodiments of this application provide an account association and identification method, apparatus, device, and storage medium. The provided account association method enables more comprehensive analysis and processing of different types of user characteristic data in each account to be associated, effectively improving the accuracy of associating different accounts of the same user. Furthermore, this application also provides an account identification method that, given multiple game accounts of the same user in a target game, can specifically address the problem of distinguishing the master-slave relationship between different game accounts of the same user. This is beneficial for creating a fairer and more just gaming environment and promoting the healthy development of the game.

[0047] To facilitate understanding of the embodiments of this application, the account association method in the account association and identification method provided in the embodiments of this application will be described in detail below.

[0048] Reference Figure 1 As shown, Figure 1 This illustration shows a flowchart of an account association method provided in an embodiment of this application. The account association method includes steps S101-S104; specifically:

[0049] S101, obtain the first user feature data of multiple accounts to be associated, perform one-hot encoding on each of the first user feature data, and obtain the first feature encoding result of each of the accounts to be associated.

[0050] S102, obtain the second user feature data of each of the accounts to be associated, cluster all the second user feature data according to the edit distance between the second user feature data of different accounts to be associated, obtain the first clustering result, and perform one-hot encoding on each cluster in the first clustering result to obtain the second feature encoding result of each account to be associated.

[0051] S103, based on the first feature encoding result and the second feature encoding result, the plurality of accounts to be associated are clustered to obtain a second clustering result.

[0052] S104, within each cluster of the second clustering result, calculate the similarity between different accounts to be associated within that cluster, and determine the target associated account from the plurality of accounts to be associated based on the similarity.

[0053] The account association method provided in this application involves: acquiring first user feature data of multiple accounts to be associated; performing one-hot encoding on each first user feature data to obtain a first feature encoding result for each account to be associated; acquiring second user feature data of each account to be associated; clustering all second user feature data based on the edit distance between the second user feature data of different accounts to be associated to obtain a first clustering result; performing one-hot encoding on each cluster in the first clustering result to obtain a second feature encoding result for each account to be associated; clustering multiple accounts to be associated based on the first and second feature encoding results to obtain a second clustering result; calculating the similarity between different accounts to be associated within each cluster in the second clustering result; and determining the target associated account from the multiple accounts to be associated based on the similarity. In this way, this application can perform more comprehensive analysis and processing of different types of user feature data in each account to be associated, effectively improving the accuracy of associating different accounts of the same user.

[0054] It should be noted that the account association method provided in this application embodiment can be applied to terminal devices or servers. This application embodiment does not limit the specific device type for the above account association method.

[0055] The following describes each step of the account association method provided in this application embodiment, taking its application to a server as an example:

[0056] S101, obtain the first user feature data of multiple accounts to be associated, perform one-hot encoding on each of the first user feature data, and obtain the first feature encoding result of each of the accounts to be associated.

[0057] Here, the aforementioned multiple accounts to be associated are different application accounts within the same application; the application can be a social application or a game application, and this application embodiment does not limit the specific application type of the aforementioned applications.

[0058] Here, the aforementioned first user feature data does not support user-defined feature data. That is, the aforementioned first user feature data belongs to feature data with small / basic differences between different accounts of the same user. Based on this, in this embodiment of the application, the aforementioned first user feature belongs to feature data that does not need to undergo clustering processing and can be directly encoded using one-hot encoding.

[0059] Specifically, the aforementioned first user characteristic data includes, but is not limited to, at least one of the following: the IP (Internet Protocol) address of the account to be associated, the MAC (Media Access Control) address of the account to be associated, the login device model (e.g., mobile phone model, computer model, etc.) of the account to be associated, and the registered mobile phone number of the account to be associated. This application embodiment does not limit the specific characteristic data content or the specific characteristic data quantity included in the aforementioned first user characteristic data.

[0060] It should be noted that, although there are various machine learning and deep learning encoding methods in natural language processing, most of them are suitable for encoding content involving syntax / semantics. However, in this embodiment, based on the purpose of converting complex first user feature data into a fixed-length feature vector (i.e., the first feature encoding result mentioned above), and since it does not involve semantic or syntactic analysis, one-hot encoding is chosen to encode the first user feature data.

[0061] S102, obtain the second user feature data of each of the accounts to be associated, cluster all the second user feature data according to the edit distance between the second user feature data of different accounts to be associated, obtain the first clustering result, and perform one-hot encoding on each cluster in the first clustering result to obtain the second feature encoding result of each account to be associated.

[0062] Here, unlike the first user feature data mentioned above, the second user feature data is used to characterize data content belonging to user-defined types. That is, for different accounts of the same user, there may be some similarity between the second user feature data within different accounts, but they will not be as similar as the first user feature data, which is basically without difference. Furthermore, since the second user feature data contains user-defined data content, the data classification of the second user feature data is more complex than that of the first user feature data. Therefore, it is necessary to first perform clustering processing on the second user feature data to obtain the basic data classification of the second user feature data, and then perform one-hot encoding on the classified second user feature data (i.e., each cluster in the first clustering result mentioned above) to improve the encoding accuracy of the second user feature data.

[0063] Specifically, based on the data characteristics of the aforementioned second user feature data, and considering that users can usually customize their account nicknames when registering an account, and that users also name their email addresses themselves, in this embodiment of the application, the aforementioned second user feature data includes at least: the user nickname of the account to be associated and / or the user email address of the account to be associated; this embodiment of the application also does not impose any limitations on the specific feature data content and the specific number of feature data included in the aforementioned second user feature data.

[0064] Based on this, when performing step S102 above, the user nicknames contained in the second user feature data can be clustered as shown in step a1 below, specifically:

[0065] Step a1: Based on the edit distance between the nicknames of different accounts to be associated, cluster the nicknames of all accounts to be associated to obtain the nickname clustering results.

[0066] When performing step S102 above, the user email addresses included in the second user feature data can be clustered as shown in step a2 below, specifically:

[0067] Step a2: Based on the edit distance between the user emails of different accounts to be associated, cluster the user emails of all accounts to be associated to obtain the user email clustering results.

[0068] Specifically, the edit distance mentioned above refers to the minimum number of edits required to convert the second user feature data of two different accounts to be associated. The larger the edit distance, the greater the difference between the second user feature data of the two accounts to be associated, and the smaller the edit distance, the smaller the difference between the second user feature data of the two accounts to be associated.

[0069] It should be noted that different editing distance calculation methods may be used when calculating the above-mentioned edit distance, resulting in different calculation results. For example, in the Levinstein distance calculation method, when calculating the minimum number of edits required to convert between the second user feature data of two different accounts to be associated, the edit operations that can be performed include: deleting, adding, and replacing a single character. In the LCS (longest common subsequence) distance calculation method, only deletion and character addition operations are allowed. Therefore, this application only needs to use a unified editing distance method to calculate the second user feature data of different accounts to be associated, ensuring the consistency of the calculation results. This application does not limit the specific calculation method of the above-mentioned edit distance in any way.

[0070] It should be noted that when performing clustering processing on all the second user feature data, any clustering algorithm can be used, such as kmeans clustering, DBSCAN (Density-Based Spatial Clustering of Applications with Noise), spectral clustering, etc. This application embodiment does not limit the specific clustering algorithm used in the above step S102.

[0071] S103, based on the first feature encoding result and the second feature encoding result, the plurality of accounts to be associated are clustered to obtain a second clustering result.

[0072] Here, based on the analysis in steps S101 and S103 above, it can be seen that the first user feature data and the second user feature data belong to different types of feature data. Therefore, there is a difference in the units of measurement between the first feature encoding result obtained based on the first user feature data and the second feature encoding result obtained based on the second user feature data. Based on this, when performing step S103, in a preferred embodiment, the first feature encoding result and the second feature encoding result of each account to be associated can be normalized first, and then the above multiple accounts to be associated can be clustered according to the normalization result to obtain the second clustering result, so as to avoid the situation where the clustering result is inaccurate due to the difference in units of measurement.

[0073] Specifically, when performing the above normalization process, the max-min (maximum and minimum values) normalization method can be used, or other types of normalization methods such as standard normalization can be used. This application embodiment does not limit the specific normalization method used in the above preferred embodiment.

[0074] Here, in the above preferred embodiment, the normalization result corresponding to each account to be associated is essentially equivalent to the account feature vector corresponding to that account. At this time, when clustering the above multiple accounts to be associated based on the normalization result, the normalization result corresponding to all accounts to be associated can be clustered directly (equivalent to clustering the account feature vectors of all accounts to be associated) to obtain the above second clustering result.

[0075] It should be noted that, similar to step S102 above, when performing clustering processing on the normalization results of all accounts to be associated, any clustering algorithm such as kmeans clustering algorithm, DBSCAN clustering algorithm, or spectral clustering algorithm can be used. This application embodiment does not limit the specific clustering algorithm used in the above preferred embodiments.

[0076] S104, within each cluster of the second clustering result, calculate the similarity between different accounts to be associated within that cluster, and determine the target associated account from the plurality of accounts to be associated based on the similarity.

[0077] Here, the identified target associated accounts are equivalent to one or more accounts belonging to the same user among the multiple accounts to be associated; this application embodiment does not limit the specific number of accounts to be associated included in the target associated accounts.

[0078] Here, based on the implementation of the aforementioned steps S101-S103, the most similar accounts to be associated among the user feature data (i.e., the first user feature data and the second user feature data) have been clustered into the same cluster in the second clustering result through clustering processing. Therefore, in step S104, it is only necessary to perform similarity calculation on the different accounts to be associated within the same cluster in each cluster of the second clustering result, thereby effectively saving the amount of data processing during similarity calculation and improving the efficiency of similarity calculation and account association.

[0079] Specifically, since the first feature encoding result and the second feature encoding result are obtained through one-hot encoding, each account to be associated can be represented by the normalization result corresponding to that account. The normalization result is represented in the form of a feature vector. Based on this, when calculating similarity in step S104, for each cluster in the second clustering result, the similarity between the normalization results corresponding to different accounts to be associated within that cluster can be directly calculated to obtain the similarity calculation result between different accounts to be associated within that cluster. Then, different accounts to be associated whose similarity calculation result within the cluster is greater than or equal to a preset similarity threshold are taken as the target associated accounts.

[0080] It should be noted that when performing similarity calculation in step S104, any similarity calculation method such as cosine similarity, Euclidean distance, or Mahalanobis distance can be used. This application embodiment does not limit the specific similarity calculation method used in step S104.

[0081] It should be noted that when using Euclidean distance, Mahalanobis distance, or other feature distance calculation methods to calculate the similarity between different accounts to be associated within the same cluster, the smaller the calculated feature distance, the higher the similarity between the two in the similarity calculation result. The specific value of the above-mentioned preset similarity threshold can be determined according to the actual account association requirements, and this application embodiment does not impose any limitations on this.

[0082] The specific implementation process of each of the above steps in the embodiments of this application will be described in detail below:

[0083] Regarding the specific implementation process of step S101 above, for the feature data contained in the first user feature data, such as the MAC address of the account to be associated, the login device model of the account to be associated, and the registered mobile phone number of the account to be associated, the corresponding feature data encoding results can be obtained by directly performing one-hot encoding.

[0084] Regarding the IP address of the account to be associated contained in the first user feature data, in a preferred embodiment, such as Figure 2 As shown, Figure 2 This illustration shows a flowchart of a method for one-hot encoding of the IP addresses of accounts to be associated, according to an embodiment of this application. The method includes steps S201-S202, specifically:

[0085] S201, obtain the IP address of each of the accounts to be associated, and convert the IP address of the account to be associated into feature data in the target format based on the location information and service provider information contained in the IP address of the account to be associated.

[0086] For example, taking the IP address of the account x1 to be associated as an example, based on the location information and service provider information contained in the IP address of the account x1 to be associated, the IP address can be converted into feature data in the following target format: China│Zhejiang Province│Hangzhou City│Binjiang District│China Telecom.

[0087] S202, perform one-hot encoding on the feature data of the target format corresponding to each account to be associated, and obtain the IP address encoding result of each account to be associated.

[0088] For example, using the above example, if the unique hot code value for China is 1, the unique hot code value for Zhejiang Province is 7 (different unique hot code values ​​for different provinces and cities), the unique hot code value for Hangzhou City is 1 (different unique hot code values ​​for different cities within each province and city), the unique hot code value for Binjiang District is 6 (different unique hot code values ​​for different districts within each prefecture-level city), and the unique hot code value for China Telecom is 3 (different unique hot code values ​​for different service providers), then by performing unique hot coding on the converted IP address, the encoding result of the above IP address can be obtained as: 17163.

[0089] Regarding the specific implementation process of step S102 above, considering that one-hot encoding is a semantically neutral encoding method (its mapping process generally involves encoding the text that appears first), when the data order of the second user feature data of different accounts to be associated is disrupted or when new second user feature data of the accounts to be associated are obtained, the output order of the clusters in the first clustering result may change. At this time, the second feature encoding result obtained in step S102 above may be inconsistent with the second feature encoding result before the change.

[0090] In an alternative implementation, to avoid the above situation, the mapping relationship of one-hot encoding can be set in advance, and one-hot encoding is no longer performed sequentially according to the output order of the clusters in the first clustering result, so that the output order of the clusters in the first clustering result will not affect the second feature encoding result.

[0091] In another optional implementation, to avoid the above situation, before each execution of the one-hot encoding step in step S102, each cluster contained in the first clustering result can be sorted according to a certain sorting rule, so that each one-hot encoding is performed on clusters with a fixed sorting method, thereby ensuring that the second feature encoding result corresponding to the same cluster is consistent each time. For example, Figure 3 As shown, Figure 3 This illustration shows a flowchart of a method for ranking clusters contained in a first clustering result according to an embodiment of this application. The method includes steps S301-S302, specifically:

[0092] S301, based on the target character of the cluster core data in each cluster in the first clustering result, sort each cluster in the first clustering result, and output the one-hot encoding result corresponding to each cluster in sequence according to the sorted result.

[0093] Here, the target characters mentioned above include: letter symbols or number symbols that appear for the first time in the cluster core data. This application embodiment does not limit the specific character type of the target characters.

[0094] For example, if the target character is a letter symbol, then according to the order of letters A(a)-Z(z), the first letter symbol appearing in the core data of each cluster is used as the representative to sort each cluster in the first clustering result. If the first clustering result contains 20 clusters, then the one-hot encoding result corresponding to the 4th cluster is 4.

[0095] S302, the one-hot encoding result corresponding to each cluster is used as the second feature encoding result of different accounts to be associated within that cluster.

[0096] It should be noted that the second user feature data located in the same cluster in the first clustering result are the most similar second user feature data. Based on the characteristic that one-hot encoding is a one-bit effective encoding, the one-hot encoding result corresponding to each cluster can be directly used as the second feature encoding result of different accounts to be associated within the same cluster.

[0097] For example, if the second user feature data includes user nicknames, and after clustering based on the edit distance between the user nicknames of different accounts to be associated, the user nicknames of accounts x1, x2, and x3 to be associated are all located in the 4th cluster, then it can be determined that the second feature encoding result of accounts x1, x2, and x3 to be associated is 4 for the corresponding user nicknames.

[0098] Regarding the account association method described in steps S101-S104 above, when multiple accounts to be associated belong to multiple different game accounts in the same target game, unlike other application accounts (e.g., social application accounts), in addition to being distinguished based on differences in user activity, different game accounts of the same user can also be distinguished and identified as primary accounts (i.e., the user's main game account in the target game) and secondary accounts (i.e., the user's sub-game accounts in the target game) based on differences in game character data (e.g., differences in game character level, differences in game character achievement completion, etc.) and user characteristic data (e.g., differences in user game level, differences in recharge and consumption frequency, etc.) between different game accounts associated with the same user.

[0099] Based on this, in an optional implementation, taking as an example that the multiple accounts to be associated obtained in step S101 belong to multiple different game accounts in the same target game, and the target associated accounts obtained in step S104 represent multiple different target game accounts associated with the same user in the target game, for each group of target associated accounts obtained, such as Figure 4 As shown, Figure 4 This illustration shows a flowchart of a method for identifying the main game account of a user in a target associated account, according to an embodiment of this application. The method includes steps S401-S403, specifically:

[0100] S401, from each of the target game accounts within the target associated account, obtain the user feature data and the character feature data of the target game character within the target game account, and perform normalization processing on the obtained user feature data and the character feature data to obtain the account feature data of each of the target game accounts.

[0101] Here, when the target game account contains only one game character, that game character is the target game character; when the target game account contains multiple game characters, they can be sorted based on each game character's character level, achievement completion rate, and other character data to select the game character most frequently used by the user in the target game account from among the multiple game characters as the target game character.

[0102] Specifically, the aforementioned user characteristic data includes, but is not limited to, user's game VIP level, total recharge spending, recharge spending frequency, and other characteristic data. This application embodiment does not limit the specific data content of the aforementioned user characteristic data.

[0103] Specifically, the aforementioned character characteristic data includes, but is not limited to, the target game character's character level, achievement completion rate, character combat power, and equipment rating. This application embodiment does not impose any limitations on the specific data content of the aforementioned character characteristic data.

[0104] It should be noted that, similar to step S103 above, in order to eliminate the impact of different dimensions of different feature data on the account score calculation results in subsequent steps, the obtained user feature data and role feature data can also be normalized first. Among them, the max-min (maximum and minimum values) normalization method can be used, or other types of normalization methods such as standard normalization can be used. This application embodiment does not limit the specific normalization method used in step S401 above.

[0105] S402, the account feature data is input into the pre-trained account score calculation model, and the account score calculation result for each target game account is output.

[0106] Here, the account score calculation result is used to characterize the degree of association between each target game account and the target user corresponding to the above-mentioned target associated account. The higher the ranking of the target game account's account score calculation result, the higher the time, money, and other costs that the target user has invested in the target game account. In other words, the target game account can be identified as the main game account used by the target user in the target game.

[0107] S403, Based on the account score calculation result of each target game account, determine the main game account of the target user in the target game.

[0108] Here, the main game account and sub-game account of the target user can be effectively distinguished based on the account score calculation result output by the model. Specifically, in an optional embodiment, determining the user's main game account in the target game based on the account score calculation result includes: determining the target game account whose account score calculation result is greater than or equal to a preset score threshold as the target user's main game account in the game, and determining other target game accounts besides the main game account as the target user's sub-game accounts in the game.

[0109] In one optional implementation, the target game account whose account score is greater than or equal to a preset score threshold can be determined directly based on the account score calculation result output by the model. The specific number of the determined main game account and sub-game account is not limited in this embodiment.

[0110] In another optional implementation, the account score calculation results of different target game accounts can be sorted in ascending order according to the account score calculation results output by the model. Then, by calculating the percentile of each target game account among all target game accounts, the target game account whose percentile ranking is above the preset ranking threshold is taken as the main game account.

[0111] The detailed calculation method for percentiles is as follows:

[0112] q_index = m × q;

[0113] Where q_index represents the percentile of the target game account at the q_index item;

[0114] m represents the total number of target game accounts;

[0115] When q_index is an integer, the qth percentile is the average of the percentiles of the q_index and (q_index+1)th percentiles; otherwise, it is rounded to the nearest integer, and the adjacent integer greater than q_index is taken as the qth percentile.

[0116] Regarding the method for identifying the main game account and sub-game account (i.e., the auxiliary game account of the same target user) described in steps S401-S403 above, it should be noted that in this embodiment, after executing steps S401-S403, that is, after determining the main game account and sub-game account of the target user in the target game, in order to maintain the fairness of the game, some account restrictions can be imposed on the sub-game accounts of different users in the game. For example, the number of times a sub-game account can participate in game tasks can be limited, or the sub-game account can be limited to participate in game battles of the same account level. The specific account restriction measures imposed on the identified sub-game accounts in the game can be determined according to the actual setting requirements of the game, and this embodiment does not impose any limitations on this.

[0117] Regarding the specific implementation process of steps S401-S403 above, the account feature data of each target game account obtained in step S401 is used as a training sample, such as... Figure 5 As shown, Figure 5 This illustration shows a flowchart of a method for training an account score calculation model according to an embodiment of this application. The method includes steps S501-S503, specifically:

[0118] S501, input the account feature data of each target game account into the initial model, and obtain the account feature encoding result corresponding to each target game account based on the target encoding function in the initial model.

[0119] Here, the initial model mentioned above can be any encoding / decoding model, and the specific type of the initial model is not limited in this application embodiment.

[0120] S502, based on the target decoding function in the initial model, decode each of the account feature encoding results to obtain the account feature decoding result corresponding to each target game account.

[0121] Here, during the training of the initial model, unsupervised learning can be used to train the initial model to encode and decode the input account feature data based on the aforementioned target encoding and decoding functions; the encoding process based on the target encoding function and the decoding process based on the target decoding function are mirror images of each other.

[0122] S503, based on the reconstruction error between the account feature decoding result corresponding to each target game account and the account feature data of each target game account, the initial model is trained unsupervised until the initial model converges, and the converged initial model is used as the account score calculation model.

[0123] Specifically, taking the target encoding function as f, the target decoding function as g, and the input account feature data as x as an example, the reconstruction error L can be calculated according to the following formula:

[0124] Reconstruction error L = argmin f，g ||xg[f(x)]||;

[0125] The argmin function is used to find the values ​​of the target encoding function f and the target decoding function g when the reconstruction error L is minimized.

[0126] The account association method provided in this application involves obtaining first user feature data for multiple accounts to be associated, performing one-hot encoding on each first user feature data to obtain a first feature encoding result for each account to be associated; obtaining second user feature data for each account to be associated, clustering all second user feature data based on the edit distance between the second user feature data of different accounts to be associated to obtain a first clustering result, and performing one-hot encoding on each cluster in the first clustering result to obtain a second feature encoding result for each account to be associated; clustering multiple accounts to be associated based on the first and second feature encoding results to obtain a second clustering result; calculating the similarity between different accounts to be associated within each cluster in the second clustering result, and determining the target associated account from multiple accounts to be associated based on the similarity. In this way, this application can perform more comprehensive analysis and processing of different types of user feature data in each account to be associated, effectively improving the accuracy of associating different accounts of the same user.

[0127] In addition to the account association method described in steps S101-S104 above, this application also provides an account identification method for multiple different game accounts of the same user in the game; wherein, referring to Figure 6 As shown, Figure 6 This illustration shows a flowchart of an account identification method provided in an embodiment of this application. The account identification method includes steps S601-S604, specifically:

[0128] S601: Obtain multiple game accounts associated with the same user in the target game.

[0129] Here, since the account association method described in steps S101-S104 above can also be used to associate different game accounts of the same user, as an optional embodiment, in step S601, the account association result of the user in the target game (i.e., the target associated account obtained in step S104 above) can be obtained as multiple game accounts associated by the user in the target game according to the account association method described in steps S101-S104 above.

[0130] In addition, in step S601, based on user registration information (e.g., mobile phone number, email address, etc.) in different game accounts in the target game, different game accounts corresponding to the same mobile phone number or the same email address can be obtained from the target game as multiple different game accounts associated with the same user in the target game; the specific method of obtaining game accounts in step S601 is not limited in this application embodiment.

[0131] It should be noted that a user can register multiple different game accounts in the target game, and the number of game accounts registered is not limited by the game type of the target game. Therefore, this application embodiment does not limit the specific game type of the target game.

[0132] S602, from each game account, obtain the user feature data and the character feature data of the target game character within the game account, and perform normalization processing on the obtained user feature data and the character feature data to obtain the account feature data of each game account.

[0133] Here, since the game account in step S602 and the target game account in step S401 are essentially multiple different game accounts associated with the same user in the game, the specific implementation of step S602 can refer to the specific implementation process of step S401 above, and the repeated parts will not be repeated here.

[0134] S603, input the account feature data into the pre-trained account score calculation model, and output the account score calculation result for each game account.

[0135] Here, the account score calculation result is used to characterize the degree of association between each of the above game accounts and the above users.

[0136] Specifically, similar to step S602, the specific implementation of step S603 can also refer to the specific implementation process of the aforementioned step S402, and the repeated parts will not be repeated here.

[0137] S604, Based on the account score calculation results of each game account, determine the user's main game account and sub-game account in the target game.

[0138] Specifically, similar to step S403, in step S604, the game account whose account score calculation result is greater than or equal to the preset score threshold can also be determined as the user's main game account in the target game, and other game accounts besides the main game account can be determined as the user's sub-game accounts in the target game; the specific implementation of the above step S604 can also refer to the specific implementation process of the aforementioned step S403, and the repeated parts will not be repeated here.

[0139] It should be noted that since the game account obtained in step S601 and the target associated account obtained in step S104 are essentially multiple different application accounts associated by the same user within the same application, regardless of whether the specific implementation of step S601 is the same as the account association method described in steps S101-S104, it does not affect the fact that steps S602-S603 can be implemented in the same way as steps S401-S403.

[0140] Regarding the account identification method described in steps S601-S604 above, the account score calculation model in step S603 can also be trained according to the training method described in steps S501-S503 above. Specifically, the training method for the account score calculation model in step S603 may include the following steps b1-b3:

[0141] Step b1: Input the account feature data of each game account into the initial model, and obtain the account feature encoding result corresponding to each game account based on the target encoding function in the initial model.

[0142] Step b2: Based on the target decoding function in the initial model, decode each account feature encoding result to obtain the account feature decoding result corresponding to each game account.

[0143] Step b3: Based on the reconstruction error between the account feature decoding result corresponding to each game account and the account feature data of each game account, perform unsupervised training on the initial model until the initial model converges, and use the converged initial model as the account score calculation model.

[0144] It should be noted that the specific implementation methods of steps b1-b3 above can refer to the specific implementation methods of steps S501-S503 above, and the repeated parts will not be repeated here.

[0145] Based on the account identification method provided in this application, multiple game accounts associated with the same user in the target game are obtained; from each game account, user feature data and character feature data of the target game character are obtained, and the obtained user feature data and character feature data are normalized to obtain account feature data for each game account; the account feature data are input into a pre-trained account score calculation model, and the account score calculation result for each game account is output; based on the account score calculation result for each game account, the user's main game account and sub-game account in the target game are determined. Thus, this application, given the knowledge of multiple game accounts of the same user in the target game, can specifically solve the problem of how to distinguish the master-slave relationship between different game accounts of the same user, which is conducive to creating a fairer and more just gaming environment and promoting the healthy development of the game.

[0146] Based on the same inventive concept, this application also provides an account association device corresponding to the above-mentioned account association method. Since the principle of the account association device in the embodiments of this application is similar to that of the above-mentioned account association method in the embodiments of this application, the implementation of the account association device can refer to the implementation of the above-mentioned account association method, and the repeated parts will not be described again.

[0147] Reference Figure 7 As shown, Figure 7 This illustration shows a structural diagram of an account association device provided in an embodiment of this application. The account association device includes:

[0148] The first acquisition module 701 is used to acquire first user feature data of multiple accounts to be associated, and perform one-hot encoding on each first user feature data to obtain the first feature encoding result of each account to be associated.

[0149] The second acquisition module 702 is used to acquire the second user feature data of each of the accounts to be associated, cluster all the second user feature data according to the edit distance between the second user feature data of different accounts to be associated, obtain the first clustering result, and perform one-hot encoding on each cluster in the first clustering result to obtain the second feature encoding result of each account to be associated.

[0150] The first processing module 703 is used to cluster the plurality of accounts to be associated based on the first feature encoding result and the second feature encoding result to obtain a second clustering result;

[0151] The association module 704 is used to calculate the similarity between different accounts to be associated within each cluster in the second clustering result, and to determine the target associated account from the plurality of accounts to be associated based on the similarity.

[0152] In an optional implementation, the first user feature data includes the IP address of the account to be associated. When acquiring the first user feature data of multiple accounts to be associated and performing one-hot encoding on each of the first user feature data, the first acquisition module 701 is specifically used for:

[0153] Obtain the IP address of each account to be associated, and convert the IP address of the account to be associated into feature data in the target format based on the location information and service provider information contained in the IP address of the account to be associated.

[0154] One-hot encoding is performed on the feature data of the target format corresponding to each account to be associated, to obtain the IP address encoding result of each account to be associated.

[0155] In one optional implementation, the first user feature data further includes at least one of the following feature data: the MAC address of the account to be associated; the login device model of the account to be associated; and the registered mobile phone number of the account to be associated.

[0156] In one optional implementation, the second user feature data includes: the user nickname of the account to be associated and / or the user email address of the account to be associated. When clustering all the second user feature data according to the edit distance between the second user feature data of different accounts to be associated to obtain a first clustering result, the second acquisition module 702 is specifically used for:

[0157] Based on the edit distance between the nicknames of different accounts to be associated, the nicknames of all accounts to be associated are clustered to obtain the nickname clustering results.

[0158] And / or,

[0159] Based on the edit distance between user emails of different accounts to be associated, the user emails of all accounts to be associated are clustered to obtain the user email clustering results.

[0160] In an optional implementation, when performing one-hot encoding on each cluster in the first clustering result to obtain the second feature encoding result for each of the accounts to be associated, the second acquisition module 702 is specifically used for:

[0161] Based on the target character of the cluster core data within each cluster in the first clustering result, each cluster in the first clustering result is sorted, and the one-hot encoding result corresponding to each cluster is output sequentially according to the sorted result; wherein, the target character includes: the first letter or number symbol appearing in the cluster core data;

[0162] The one-hot encoding result corresponding to each cluster is used as the second feature encoding result of different accounts to be associated within that cluster.

[0163] In one optional implementation, the first processing module 703 is specifically used for:

[0164] The first feature encoding result and the second feature encoding result of each of the accounts to be associated are normalized, and the multiple accounts to be associated are clustered according to the normalization result to obtain the second clustering result;

[0165] When calculating the similarity between different accounts to be associated within the cluster, the association module 704 is specifically used for:

[0166] Calculate the similarity between the normalized results corresponding to different accounts to be associated within the cluster.

[0167] In one optional implementation, the plurality of accounts to be associated includes a plurality of different game accounts in the same target game, and the target associated accounts represent a plurality of different target game accounts associated by the same user in the target game.

[0168] In one optional implementation, the account association device further includes:

[0169] The third acquisition module is used to acquire user feature data and character feature data of target game characters from each target game account in the target associated account, and to normalize the acquired user feature data and character feature data to obtain account feature data of each target game account.

[0170] The second processing module is used to input the account feature data into a pre-trained account score calculation model and output the account score calculation result for each target game account; wherein, the account score calculation result is used to characterize the degree of association between each target game account and the target user corresponding to the target associated account;

[0171] The determination module is used to determine the main game account of the target user in the target game based on the account score calculation result of each target game account.

[0172] In one optional implementation, the second processing module is used to train the account score calculation model using the following method:

[0173] The account feature data of each target game account is input into the initial model, and the account feature encoding result corresponding to each target game account is obtained based on the target encoding function in the initial model.

[0174] Based on the target decoding function in the initial model, each account feature encoding result is decoded to obtain the account feature decoding result corresponding to each target game account.

[0175] Based on the reconstruction error between the account feature decoding result corresponding to each target game account and the account feature data of each target game account, the initial model is trained unsupervised until the initial model converges, and the converged initial model is used as the account score calculation model.

[0176] The account association device provided in this application provides the following steps: First user feature data of multiple accounts to be associated are obtained; each first user feature data is one-hot encoded to obtain a first feature encoding result for each account to be associated; second user feature data of each account to be associated are obtained; based on the edit distance between the second user feature data of different accounts to be associated, all second user feature data are clustered to obtain a first clustering result; each cluster in the first clustering result is one-hot encoded to obtain a second feature encoding result for each account to be associated; based on the first and second feature encoding results, multiple accounts to be associated are clustered to obtain a second clustering result; within each cluster in the second clustering result, the similarity between different accounts to be associated within that cluster is calculated, and the target associated account is determined from the multiple accounts to be associated based on the similarity. In this way, this application can perform more comprehensive analysis and processing of different types of user feature data in each account to be associated, effectively improving the accuracy of associating different accounts of the same user.

[0177] Based on the same inventive concept, this application also provides an account recognition device corresponding to the above-mentioned account recognition method. Since the principle of the account recognition device in the embodiments of this application is similar to that of the above-mentioned account recognition method in the embodiments of this application, the implementation of the account recognition device can refer to the implementation of the above-mentioned account recognition method, and the repeated parts will not be described again.

[0178] Reference Figure 8 As shown, Figure 8 This illustration shows a structural diagram of an account recognition device provided in an embodiment of this application. The account recognition device includes:

[0179] Account Acquisition Module 801 is used to acquire multiple game accounts associated with the same user in the target game;

[0180] The data processing module 802 is used to obtain user feature data and target game character feature data from each game account, and to normalize the obtained user feature data and character feature data to obtain account feature data for each game account.

[0181] The score determination module 803 is used to input the account feature data into a pre-trained account score calculation model and output the account score calculation result for each game account; wherein, the account score calculation result is used to characterize the degree of association between each game account and the user;

[0182] The account identification module 804 is used to determine the user's main game account and sub-game account in the target game based on the account score calculation result of each game account.

[0183] In an optional implementation, the score determination module 803 is further configured to train the account score calculation model using the following method:

[0184] The account feature data of each game account is input into the initial model, and the account feature encoding result corresponding to each game account is obtained based on the target encoding function in the initial model.

[0185] Based on the target decoding function in the initial model, the encoding result of each account feature is decoded to obtain the account feature decoding result corresponding to each game account.

[0186] Based on the reconstruction error between the account feature decoding result corresponding to each game account and the account feature data of each game account, the initial model is trained unsupervised until the initial model converges, and the converged initial model is used as the account score calculation model.

[0187] In one alternative implementation, the account recognition module 804 is specifically used for:

[0188] Game accounts whose account score calculation result is greater than or equal to a preset score threshold are identified as the main game account, and other game accounts besides the main game account are identified as the sub-game accounts.

[0189] Based on the account identification device provided in this application embodiment, multiple game accounts associated with the same user in the target game are obtained; from each game account, user feature data and character feature data of the target game character are obtained, and the obtained user feature data and character feature data are normalized to obtain account feature data for each game account; the account feature data are input into a pre-trained account score calculation model, and the account score calculation result for each game account is output; based on the account score calculation result for each game account, the user's main game account and sub-game account in the target game are determined. Thus, this application, given the knowledge of multiple game accounts of the same user in the target game, can specifically solve the problem of how to distinguish the master-slave relationship between different game accounts of the same user, which is conducive to creating a fairer and more just gaming environment and promoting the healthy development of the game.

[0190] Corresponding to the account association method provided in the embodiments of this application, based on the same inventive concept, Figure 9 A schematic diagram of the structure of an electronic device 900 provided in this application embodiment includes: a processor 901, a memory 902, and a bus 903. The memory 902 stores machine-readable instructions executable by the processor 901. When the electronic device 900 runs an account association method as described in the embodiment, the processor 901 communicates with the memory 902 via the bus 903. The processor 901 executes the machine-readable instructions, wherein the processor 901 executes the following steps when executing the machine-readable instructions:

[0191] Obtain first user feature data of multiple accounts to be associated, perform one-hot encoding on each first user feature data, and obtain the first feature encoding result of each account to be associated.

[0192] Obtain the second user feature data of each of the accounts to be associated; cluster all the second user feature data according to the edit distance between the second user feature data of different accounts to be associated to obtain the first clustering result; and perform one-hot encoding on each cluster in the first clustering result to obtain the second feature encoding result of each account to be associated.

[0193] Based on the first feature encoding result and the second feature encoding result, the plurality of accounts to be associated are clustered to obtain a second clustering result;

[0194] Within each cluster of the second clustering result, the similarity between different accounts to be associated within that cluster is calculated, and the target associated account is determined from the plurality of accounts to be associated based on the similarity.

[0195] In an optional implementation, the first user feature data includes the IP address of the account to be associated. When acquiring the first user feature data of multiple accounts to be associated and performing one-hot encoding on each of the first user feature data, the processor 901 is specifically used for:

[0196] Obtain the IP address of each account to be associated, and convert the IP address of the account to be associated into feature data in the target format based on the location information and service provider information contained in the IP address of the account to be associated.

[0197] One-hot encoding is performed on the feature data of the target format corresponding to each account to be associated, to obtain the IP address encoding result of each account to be associated.

[0198] In one optional implementation, the first user feature data further includes at least one of the following feature data: the MAC address of the account to be associated; the login device model of the account to be associated; and the registered mobile phone number of the account to be associated.

[0199] In an optional implementation, the second user feature data includes the user nickname of the account to be associated and / or the user email address of the account to be associated. When clustering all the second user feature data according to the edit distance between the second user feature data of different accounts to be associated to obtain a first clustering result, the processor 901 is specifically used to:

[0200] Based on the edit distance between the nicknames of different accounts to be associated, the nicknames of all accounts to be associated are clustered to obtain the nickname clustering results.

[0201] And / or,

[0202] Based on the edit distance between user emails of different accounts to be associated, the user emails of all accounts to be associated are clustered to obtain the user email clustering results.

[0203] In an optional implementation, when performing one-hot encoding on each cluster in the first clustering result to obtain the second feature encoding result for each of the accounts to be associated, the processor 901 is specifically used for:

[0204] Based on the target character of the cluster core data within each cluster in the first clustering result, each cluster in the first clustering result is sorted, and the one-hot encoding result corresponding to each cluster is output sequentially according to the sorted result; wherein, the target character includes: the first letter or number symbol appearing in the cluster core data;

[0205] The one-hot encoding result corresponding to each cluster is used as the second feature encoding result of different accounts to be associated within that cluster.

[0206] In an optional implementation, when clustering the plurality of accounts to be associated based on the first feature encoding result and the second feature encoding result to obtain a second clustering result, the processor 901 is specifically used to:

[0207] The first feature encoding result and the second feature encoding result of each of the accounts to be associated are normalized, and the multiple accounts to be associated are clustered according to the normalization result to obtain the second clustering result;

[0208] When calculating the similarity between different accounts to be associated within the cluster, processor 901 is specifically used for:

[0209] Calculate the similarity between the normalized results corresponding to different accounts to be associated within the cluster.

[0210] In an optional implementation, when determining the target associated account from the plurality of accounts to be associated based on the similarity, the processor 901 is specifically configured to:

[0211] Different accounts to be associated within the cluster whose similarity calculation results are greater than or equal to a preset similarity threshold are taken as the target associated accounts.

[0212] In one optional implementation, the plurality of accounts to be associated includes a plurality of different game accounts in the same target game, and the target associated accounts represent a plurality of different target game accounts associated by the same user in the target game.

[0213] In an optional implementation, after obtaining the target associated account, the processor 901 is further configured to:

[0214] From each of the target game accounts within the target associated account, obtain the user feature data and the character feature data of the target game character within that target game account, and perform normalization processing on the obtained user feature data and the character feature data to obtain the account feature data of each target game account.

[0215] The account feature data is input into a pre-trained account score calculation model, and the output is the account score calculation result for each target game account; wherein, the account score calculation result is used to characterize the degree of association between each target game account and the target user corresponding to the target associated account;

[0216] Based on the account score calculation results of each target game account, the main game account of the target user in the target game is determined.

[0217] In an optional implementation, the processor 901 is further configured to train the account score calculation model using the following method:

[0218] The account feature data of each target game account is input into the initial model, and the account feature encoding result corresponding to each target game account is obtained based on the target encoding function in the initial model.

[0219] Based on the target decoding function in the initial model, each account feature encoding result is decoded to obtain the account feature decoding result corresponding to each target game account.

[0220] Based on the reconstruction error between the account feature decoding result corresponding to each target game account and the account feature data of each target game account, the initial model is trained unsupervised until the initial model converges, and the converged initial model is used as the account score calculation model.

[0221] The electronic device 900 provided in this application embodiment acquires first user feature data of multiple accounts to be associated, performs one-hot encoding on each first user feature data, and obtains a first feature encoding result for each account to be associated; acquires second user feature data of each account to be associated, clusters all second user feature data according to the edit distance between the second user feature data of different accounts to be associated, obtains a first clustering result, and performs one-hot encoding on each cluster in the first clustering result, obtaining a second feature encoding result for each account to be associated; clusters multiple accounts to be associated according to the first feature encoding result and the second feature encoding result, obtains a second clustering result; within each cluster in the second clustering result, calculates the similarity between different accounts to be associated within that cluster, and determines the target associated account from multiple accounts to be associated based on the similarity. In this way, this application can perform more comprehensive analysis and processing of different types of user feature data in each account to be associated, effectively improving the accuracy of associating different accounts of the same user.

[0222] Corresponding to the account association method provided in the embodiments of this application, the embodiments of this application also provide a computer-readable storage medium storing a computer program, which is executed by a processor, wherein the processor performs the following steps:

[0223] Obtain first user feature data of multiple accounts to be associated, perform one-hot encoding on each first user feature data, and obtain the first feature encoding result of each account to be associated.

[0224] Obtain the second user feature data of each of the accounts to be associated; cluster all the second user feature data according to the edit distance between the second user feature data of different accounts to be associated to obtain the first clustering result; and perform one-hot encoding on each cluster in the first clustering result to obtain the second feature encoding result of each account to be associated.

[0225] Based on the first feature encoding result and the second feature encoding result, the plurality of accounts to be associated are clustered to obtain a second clustering result;

[0226] Within each cluster of the second clustering result, the similarity between different accounts to be associated within that cluster is calculated, and the target associated account is determined from the plurality of accounts to be associated based on the similarity.

[0227] In an optional implementation, the first user feature data includes the IP address of the account to be associated. When acquiring the first user feature data of multiple accounts to be associated and performing one-hot encoding on each of the first user feature data, the processor is specifically configured to:

[0228] Obtain the IP address of each account to be associated, and convert the IP address of the account to be associated into feature data in the target format based on the location information and service provider information contained in the IP address of the account to be associated.

[0229] One-hot encoding is performed on the feature data of the target format corresponding to each account to be associated, to obtain the IP address encoding result of each account to be associated.

[0230] In one optional implementation, the first user feature data further includes at least one of the following feature data: the MAC address of the account to be associated; the login device model of the account to be associated; and the registered mobile phone number of the account to be associated.

[0231] In one optional implementation, the second user feature data includes the user nickname of the account to be associated and / or the user email address of the account to be associated. When clustering all the second user feature data based on the edit distance between the second user feature data of different accounts to be associated to obtain a first clustering result, the processor is specifically used for:

[0232] Based on the edit distance between the nicknames of different accounts to be associated, the nicknames of all accounts to be associated are clustered to obtain the nickname clustering results.

[0233] And / or,

[0234] Based on the edit distance between user emails of different accounts to be associated, the user emails of all accounts to be associated are clustered to obtain the user email clustering results.

[0235] In an optional implementation, when performing one-hot encoding on each cluster in the first clustering result to obtain the second feature encoding result for each of the accounts to be associated, the processor is specifically used for:

[0236] Based on the target character of the cluster core data within each cluster in the first clustering result, each cluster in the first clustering result is sorted, and the one-hot encoding result corresponding to each cluster is output sequentially according to the sorted result; wherein, the target character includes: the first letter or number symbol appearing in the cluster core data;

[0237] The one-hot encoding result corresponding to each cluster is used as the second feature encoding result of different accounts to be associated within that cluster.

[0238] In an optional implementation, when clustering the plurality of accounts to be associated based on the first feature encoding result and the second feature encoding result to obtain a second clustering result, the processor is specifically used to:

[0239] The first feature encoding result and the second feature encoding result of each of the accounts to be associated are normalized, and the multiple accounts to be associated are clustered according to the normalization result to obtain the second clustering result;

[0240] When calculating the similarity between different accounts to be associated within the cluster, the processor is specifically used to:

[0241] Calculate the similarity between the normalized results corresponding to different accounts to be associated within the cluster.

[0242] In an optional implementation, when determining the target associated account from the plurality of accounts to be associated based on the similarity, the processor is specifically configured to:

[0243] Different accounts to be associated within the cluster whose similarity calculation results are greater than or equal to a preset similarity threshold are taken as the target associated accounts.

[0244] In one optional implementation, the plurality of accounts to be associated includes a plurality of different game accounts in the same target game, and the target associated accounts represent a plurality of different target game accounts associated by the same user in the target game.

[0245] In one optional implementation, after obtaining the target associated account, the processor is further configured to:

[0246] From each of the target game accounts within the target associated account, obtain the user feature data and the character feature data of the target game character within that target game account, and perform normalization processing on the obtained user feature data and the character feature data to obtain the account feature data of each target game account.

[0247] The account feature data is input into a pre-trained account score calculation model, and the output is the account score calculation result for each target game account; wherein, the account score calculation result is used to characterize the degree of association between each target game account and the target user corresponding to the target associated account;

[0248] Based on the account score calculation results of each target game account, the main game account of the target user in the target game is determined.

[0249] In an optional implementation, the processor is further configured to train the account score calculation model using the following method:

[0250] The account feature data of each target game account is input into the initial model, and the account feature encoding result corresponding to each target game account is obtained based on the target encoding function in the initial model.

[0251] Based on the target decoding function in the initial model, each account feature encoding result is decoded to obtain the account feature decoding result corresponding to each target game account.

[0252] Based on the reconstruction error between the account feature decoding result corresponding to each target game account and the account feature data of each target game account, the initial model is trained unsupervised until the initial model converges, and the converged initial model is used as the account score calculation model.

[0253] Using the computer-readable storage medium provided in this application embodiment, first user feature data of multiple accounts to be associated are obtained. Each first user feature data is one-hot encoded to obtain a first feature encoding result for each account to be associated. Second user feature data of each account to be associated are obtained. Based on the edit distance between the second user feature data of different accounts to be associated, all second user feature data are clustered to obtain a first clustering result. Each cluster in the first clustering result is one-hot encoded to obtain a second feature encoding result for each account to be associated. Based on the first and second feature encoding results, multiple accounts to be associated are clustered to obtain a second clustering result. Within each cluster in the second clustering result, the similarity between different accounts to be associated within that cluster is calculated, and the target associated account is determined from the multiple accounts to be associated based on the similarity. Thus, this application can perform more comprehensive analysis and processing of different types of user feature data in each account to be associated, effectively improving the accuracy of associating different accounts of the same user.

[0254] In this embodiment, the computer program, when run by the processor, can also execute other machine-readable instructions to perform the account association methods described in other embodiments. For details on the specific execution steps and principles, please refer to the description of the embodiments, which will not be repeated here.

[0255] Corresponding to the account identification method provided in the embodiments of this application, based on the same inventive concept, Figure 10 A schematic diagram of another electronic device 1000 provided in this application embodiment includes: a processor 1001, a memory 1002, and a bus 1003. The memory 1002 stores machine-readable instructions executable by the processor 1001. When the electronic device 1000 runs an account recognition method as described in the embodiment, the processor 1001 and the memory 1002 communicate via the bus 1003. The processor 1001 executes the machine-readable instructions, wherein the processor 1001 executes the machine-readable instructions to implement the following steps:

[0256] Obtain multiple game accounts associated with the same user in the target game;

[0257] From each game account, user feature data and target game character feature data within the game account are obtained, and the obtained user feature data and target game character feature data are normalized to obtain account feature data for each game account.

[0258] The account feature data is input into a pre-trained account score calculation model, and the output is the account score calculation result for each game account; wherein, the account score calculation result is used to characterize the degree of association between each game account and the user;

[0259] Based on the account score calculation results of each game account, the user's main game account and sub-game account in the target game are determined.

[0260] In an optional implementation, the processor 1001 is further configured to train the account score calculation model using the following method:

[0261] The account feature data of each game account is input into the initial model, and the account feature encoding result corresponding to each game account is obtained based on the target encoding function in the initial model.

[0262] Based on the target decoding function in the initial model, the encoding result of each account feature is decoded to obtain the account feature decoding result corresponding to each game account.

[0263] Based on the reconstruction error between the account feature decoding result corresponding to each game account and the account feature data of each game account, the initial model is trained unsupervised until the initial model converges, and the converged initial model is used as the account score calculation model.

[0264] In an optional implementation, when determining the user's main game account and sub-game account in the target game based on the account score calculation result of each game account, the processor 1001 is configured to:

[0265] Game accounts whose account score calculation result is greater than or equal to a preset score threshold are identified as the main game account, and other game accounts besides the main game account are identified as the sub-game accounts.

[0266] Based on the electronic device 1000 provided in this application embodiment, multiple game accounts associated with the same user in the target game are obtained; from each game account, user feature data and character feature data of the target game character are obtained, and the obtained user feature data and character feature data are normalized to obtain account feature data for each game account; the account feature data are input into a pre-trained account score calculation model, and the account score calculation result for each game account is output; based on the account score calculation result for each game account, the user's main game account and sub-game account in the target game are determined. Thus, given that multiple game accounts of the same user in the target game are known, this application can specifically solve the problem of how to distinguish the master-slave relationship between different game accounts of the same user, which is conducive to creating a fairer and more just gaming environment and promoting the healthy development of the game.

[0267] Corresponding to the account identification method provided in the embodiments of this application, the embodiments of this application also provide a computer-readable storage medium storing a computer program, which is executed by a processor, wherein the processor performs the following steps:

[0268] Obtain multiple game accounts associated with the same user in the target game;

[0269] From each game account, user feature data and target game character feature data within the game account are obtained, and the obtained user feature data and target game character feature data are normalized to obtain account feature data for each game account.

[0270] The account feature data is input into a pre-trained account score calculation model, and the output is the account score calculation result for each game account; wherein, the account score calculation result is used to characterize the degree of association between each game account and the user;

[0271] Based on the account score calculation results of each game account, the user's main game account and sub-game account in the target game are determined.

[0272] In an optional implementation, the processor is further configured to train the account score calculation model using the following method:

[0273] The account feature data of each game account is input into the initial model, and the account feature encoding result corresponding to each game account is obtained based on the target encoding function in the initial model.

[0274] Based on the target decoding function in the initial model, the encoding result of each account feature is decoded to obtain the account feature decoding result corresponding to each game account.

[0275] Based on the reconstruction error between the account feature decoding result corresponding to each game account and the account feature data of each game account, the initial model is trained unsupervised until the initial model converges, and the converged initial model is used as the account score calculation model.

[0276] In one optional implementation, when determining the user's main game account and sub-game account in the target game based on the account score calculation result of each game account, the processor is configured to:

[0277] Game accounts whose account score calculation result is greater than or equal to a preset score threshold are identified as the main game account, and other game accounts besides the main game account are identified as the sub-game accounts.

[0278] Based on the computer-readable storage medium provided in the embodiments of this application, multiple game accounts associated with the same user in the target game are obtained; from each game account, user feature data and character feature data of the target game character are obtained, and the obtained user feature data and character feature data are normalized to obtain account feature data for each game account; the account feature data are input into a pre-trained account score calculation model, and the account score calculation result for each game account is output; based on the account score calculation result for each game account, the user's main game account and sub-game account in the target game are determined. Thus, given that the same user has multiple game accounts in the target game, this application can specifically solve the problem of how to distinguish the master-slave relationship between different game accounts of the same user, which is conducive to creating a fairer and more just gaming environment and promoting the healthy development of the game.

[0279] In this embodiment, the computer program, when run by the processor, can also execute other machine-readable instructions to perform the account identification methods described in other embodiments. For details on the specific execution steps and principles, please refer to the description of the embodiments, which will not be repeated here.

[0280] In the embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. The apparatus embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. Furthermore, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Additionally, the displayed or discussed mutual couplings, direct couplings, or communication connections may be through some communication interfaces; indirect couplings or communication connections between devices or units may be electrical, mechanical, or other forms.

[0281] 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.

[0282] In addition, the functional units in the embodiments provided in 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.

[0283] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0284] 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. In addition, the terms "first", "second", "third", etc. are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0285] Finally, it should be noted that the above-described embodiments are merely specific implementations of this application, used to illustrate the technical solutions of this application, and not to limit them. The protection scope of this application is not limited thereto. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments, or make equivalent substitutions for some of the technical features, within the scope of the technology disclosed in this application; and these modifications, changes, or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application. All should be covered within the protection scope of this application. Therefore, the protection scope of this application should be determined by the protection scope of the claims.

Claims

1. A method for account association, characterized in that, The account association method includes: First user feature data of multiple accounts to be associated are obtained, and one-hot encoding is performed on each first user feature data to obtain the first feature encoding result of each account to be associated; wherein, the first user feature data represents data content that does not support user-defined feature data. The second user feature data of each of the accounts to be associated is obtained. Based on the edit distance between the second user feature data of different accounts to be associated, all the second user feature data are clustered to obtain a first clustering result. Each cluster in the first clustering result is one-hot encoded to obtain the second feature encoding result of each account to be associated. The second user feature data represents feature data whose data content belongs to a user-defined type. Based on the first feature encoding result and the second feature encoding result, the plurality of accounts to be associated are clustered to obtain a second clustering result; Within each cluster of the second clustering result, the similarity between different accounts to be associated within that cluster is calculated, and the target associated account is determined from the plurality of accounts to be associated based on the similarity.

2. The account association method according to claim 1, characterized in that, The first user feature data includes the IP address of the account to be associated. The step of obtaining first user feature data for multiple accounts to be associated, and performing one-hot encoding on each of the first user feature data, includes: Obtain the IP address of each account to be associated, and convert the IP address of the account to be associated into feature data in the target format based on the location information and service provider information contained in the IP address of the account to be associated. One-hot encoding is performed on the feature data of the target format corresponding to each account to be associated, to obtain the IP address encoding result of each account to be associated.

3. The account association method according to claim 1, characterized in that, The first user feature data also includes at least one of the following feature data: the MAC address of the account to be associated; the login device model of the account to be associated; and the registered mobile phone number of the account to be associated.

4. The account association method according to claim 1, characterized in that, The second user feature data includes: the user nickname of the account to be associated and / or the user email address of the account to be associated; the step of clustering all the second user feature data according to the edit distance between the second user feature data of different accounts to be associated, to obtain a first clustering result, includes: Cluster the user nicknames of all accounts to be associated based on the edit distance between the user nicknames of different accounts to be associated, and obtain the user nickname clustering results; and / or, cluster the user emails of all accounts to be associated based on the edit distance between the user emails of different accounts to be associated, and obtain the user email clustering results.

5. The account association method according to claim 1, characterized in that, The step of performing one-hot encoding on each cluster in the first clustering result to obtain the second feature encoding result for each of the accounts to be associated includes: Based on the target character of the cluster core data within each cluster in the first clustering result, each cluster in the first clustering result is sorted, and the one-hot encoding result corresponding to each cluster is output sequentially according to the sorted result; wherein, the target character includes: the first letter or number symbol appearing in the cluster core data; The one-hot encoding result corresponding to each cluster is used as the second feature encoding result of different accounts to be associated within that cluster.

6. The account association method according to claim 1, characterized in that, The step of clustering the multiple accounts to be associated based on the first feature encoding result and the second feature encoding result to obtain a second clustering result includes: The first feature encoding result and the second feature encoding result of each of the accounts to be associated are normalized, and the multiple accounts to be associated are clustered according to the normalization result to obtain the second clustering result; The calculation of the similarity between different accounts to be associated within the cluster includes: Calculate the similarity between the normalized results corresponding to different accounts to be associated within the cluster.

7. The account association method according to claim 1, characterized in that, The step of determining the target associated account from the plurality of accounts to be associated based on the similarity includes: Different accounts to be associated within the cluster whose similarity calculation results are greater than or equal to a preset similarity threshold are taken as the target associated accounts.

8. The account association method according to claim 1, characterized in that, The multiple accounts to be associated include multiple different game accounts in the same target game, and the target associated accounts represent multiple different target game accounts associated with the same user in the target game.

9. The account association method according to claim 8, characterized in that, After obtaining the target associated account, the account association method further includes: From each of the target game accounts within the target associated account, obtain the user feature data and the character feature data of the target game character within that target game account, and perform normalization processing on the obtained user feature data and the character feature data to obtain the account feature data of each target game account. The account feature data is input into a pre-trained account score calculation model, and the output is the account score calculation result for each target game account; wherein, the account score calculation result is used to characterize the degree of association between each target game account and the target user corresponding to the target associated account; Based on the account score calculation results of each target game account, the main game account of the target user in the target game is determined.

10. The account association method according to claim 9, characterized in that, The account score calculation model was trained using the following method: The account feature data of each target game account is input into the initial model, and the account feature encoding result corresponding to each target game account is obtained based on the target encoding function in the initial model. Based on the target decoding function in the initial model, each account feature encoding result is decoded to obtain the account feature decoding result corresponding to each target game account. Based on the reconstruction error between the account feature decoding result corresponding to each target game account and the account feature data of each target game account, the initial model is trained unsupervised until the initial model converges, and the converged initial model is used as the account score calculation model.

11. An account identification method, characterized in that, The account identification method includes: Obtain multiple game accounts associated with the same user in the target game; From each game account, obtain the user feature data and the character feature data of the target game character within the game account, and perform normalization processing on the obtained user feature data and the character feature data to obtain the account feature data of each game account. The account feature data is input into a pre-trained account score calculation model, and the output is the account score calculation result for each game account; wherein, the account score calculation result is used to characterize the degree of association between each game account and the user; Based on the account score calculation results of each game account, the user's main game account and sub-game account in the target game are determined.

12. The account identification method according to claim 11, characterized in that, The account score calculation model was trained using the following method: The account feature data of each game account is input into the initial model, and the account feature encoding result corresponding to each game account is obtained based on the target encoding function in the initial model. Based on the target decoding function in the initial model, the encoding result of each account feature is decoded to obtain the account feature decoding result corresponding to each game account. Based on the reconstruction error between the account feature decoding result corresponding to each game account and the account feature data of each game account, the initial model is trained unsupervised until the initial model converges, and the converged initial model is used as the account score calculation model.

13. The account identification method according to claim 11, characterized in that, The step of determining the user's main game account and sub-game account in the target game based on the account score calculation result of each game account includes: Game accounts whose account score calculation result is greater than or equal to a preset score threshold are identified as the main game account, and other game accounts besides the main game account are identified as the sub-game accounts.

14. An account association device, characterized in that, The account association device includes: The first acquisition module is used to acquire first user feature data of multiple accounts to be associated, perform one-hot encoding on each first user feature data, and obtain the first feature encoding result of each account to be associated; wherein, the first user feature data represents data content that does not support user-defined feature data. The second acquisition module is used to acquire the second user feature data of each of the accounts to be associated, and to cluster all the second user feature data according to the edit distance between the second user feature data of different accounts to be associated, to obtain a first clustering result, and to perform one-hot encoding on each cluster in the first clustering result to obtain the second feature encoding result of each account to be associated; wherein, the second user feature data represents feature data whose data content belongs to the user-defined type. The first processing module is used to cluster the plurality of accounts to be associated based on the first feature encoding result and the second feature encoding result to obtain a second clustering result; The association module is used to calculate the similarity between different accounts to be associated within each cluster in the second clustering result, and to determine the target associated account from the plurality of accounts to be associated based on the similarity.

15. An account recognition device, characterized in that, The account identification device includes: The account acquisition module is used to acquire multiple game accounts associated with the same user in the target game; The data processing module is used to obtain user feature data and target game character feature data from each game account, and to normalize the obtained user feature data and character feature data to obtain account feature data for each game account. The score determination module is used to input the account feature data into a pre-trained account score calculation model and output the account score calculation result for each game account; wherein, the account score calculation result is used to characterize the degree of association between each game account and the user; The account identification module is used to determine the user's main game account and sub-game account in the target game based on the account score calculation results of each game account.

16. An electronic device, characterized in that, include: The device includes a processor, a memory, and a bus. The memory stores machine-readable instructions executable by the processor. When the electronic device is running, the processor communicates with the memory via the bus. When the machine-readable instructions are executed by the processor, they perform the steps of the account association method as described in any one of claims 1 to 10, or the steps of the account identification method as described in any one of claims 11 to 13.

17. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, performs the steps of the account association method as described in any one of claims 1 to 10, or the steps of the account identification method as described in any one of claims 11 to 13.

Images