Account relationship determination method and apparatus, server, and storage medium

Abstract

Embodiments of the present application disclose a kind of account relationship determination method, device, server and storage medium, belong to computer technology field.The method comprises: determining the first associated account and the second associated account corresponding to target account from account relationship network;According to the first relationship type between the first associated account and target account, determine the first probability distribution of first associated account;Based on the first probability distribution, and the historical interaction record between the first associated account and the second associated account, determine the second probability distribution of second associated account;According to the second probability distribution, determine the second relationship type between the second associated account and target account.The relationship type between other each account is determined by the higher interaction and the account relationship of easy to determine in the embodiments of the present application, without obtaining the detailed information of all accounts in account relationship network, reduce the cost of account relationship determination, improve the efficiency and accuracy of determining account relationship.

Description

TECHNICAL FIELD

[0001] Embodiments of the present application relate to the technical field of computer, in particular to a method and device for determining account relationship, a server and a storage medium. BACKGROUND

[0002] With the popularization of computer technology, people interact with others through the Internet. Since the interaction behaviors between users of different types of social relationships are different, developers can obtain the real social relationship between the users corresponding to the target account and the associated account by analyzing the interaction behaviors, so as to provide personalized services to different users, collect information, efficiently investigate abnormal accounts, and improve network security, etc.

[0003] In related technologies, the method for obtaining the relationship type between users is to determine the relationship type between the two according to the detailed personal information corresponding to the accounts of the two interactive parties, or to cluster the interactive associated accounts with high feature similarity by extracting the interaction behavior features between the target account and the interactive associated accounts.

[0004] However, if the relationship type is determined according to detailed information, the relationship type between the account and other accounts cannot be determined for accounts with incomplete information or low interaction frequency. If clustering is performed according to interaction behavior features, the relationship between each pair of accounts cannot be determined, that is, the related technology cannot directly determine the relationship type between each pair of accounts according to the interaction behavior, and the efficiency and accuracy of relationship classification are low. SUMMARY

[0005] Embodiments of the present application provide a method and device for determining account relationship, a server and a storage medium, which can determine the relationship type between each pair of interactive accounts without obtaining detailed information of all accounts in the account relationship network, thereby reducing the cost of determining account relationship and improving the efficiency and accuracy of determining account relationship. The technical solution is as follows:

[0006] In one aspect, the present application provides a method for determining account relationship, which comprises:

[0007] determining a first associated account and a second associated account corresponding to a target account from an account relationship network, the interaction between the first associated account and the target account being higher than the interaction between the second associated account and the target account, the account relationship network being a relationship network formed based on historical interaction records between accounts;

[0008] determining a first probability distribution of the first associated account according to a first relationship type between the first associated account and the target account, the first probability distribution being used to represent the probability distribution of the relationship type between the first associated account and the target account, and the relationship type comprising at least two types;

[0009] determine a second probability distribution of the second associated account based on the first probability distribution and the historical interaction records between the first associated account and the second associated account, the second probability distribution being used to represent a probability distribution condition of a relationship type between the second associated account and the target account;

[0010] determine a second relationship type between the second associated account and the target account according to the second probability distribution.

[0011] In another aspect, an embodiment of the present application provides a device for determining an account relationship, the device comprising:

[0012] a first determining module configured to determine a first associated account and a second associated account corresponding to a target account from an account relationship network, the interaction between the first associated account and the target account being higher than the interaction between the second associated account and the target account, the account relationship network being a relationship network formed based on historical interaction records between accounts;

[0013] a second determining module configured to determine a first probability distribution of the first associated account according to a first relationship type between the first associated account and the target account, the first probability distribution being used to represent a probability distribution condition of the relationship type between the first associated account and the target account, the relationship type comprising at least two types;

[0014] a third determining module configured to determine a second probability distribution of the second associated account based on the first probability distribution and the historical interaction records between the first associated account and the second associated account, the second probability distribution being used to represent a probability distribution condition of a relationship type between the second associated account and the target account;

[0015] a fourth determining module configured to determine a second relationship type between the second associated account and the target account according to the second probability distribution.

[0016] Optionally, the third determining module comprises:

[0017] an initialization unit configured to initialize the second probability distribution according to the first probability distribution in response to the historical interaction records existing between the second associated account and the first associated account;

[0018] an updating unit configured to iteratively update the second probability distribution based on the historical interaction records between the first associated account and the second associated account and the historical interaction records between the second associated accounts.

[0019] Optionally, the initialization unit is further configured to:

[0020] determine a first statistical variable corresponding to the first related account based on the first probability distribution, different statistical variables corresponding to different relationship types;

[0021] transmit the first statistical variable to the second related account;

[0022] initialize the second probability distribution according to the first statistical variable received by the second related account, a probability of a relationship type in the second probability distribution being a ratio of a number of the first statistical variable corresponding to the relationship type to a total number of the first statistical variables.

[0023] Optionally, the updating unit is further configured to:

[0024] determine the first statistical variable based on the first probability distribution, and determine a second statistical variable corresponding to the second related account based on the second probability distribution, the first statistical variable and the second statistical variable corresponding to a same relationship type being same;

[0025] transmit the first statistical variable to the second related account, and transmit the second statistical variable to another second related account having the historical interaction record with the second related account;

[0026] update the second probability distribution according to the first statistical variable and the second statistical variable received by the second related account.

[0027] Optionally, the fourth determination module comprises:

[0028] a first determination unit, configured to determine the second relationship type between the second related account and the target account according to the second probability distribution in response to the second probability distribution satisfying a convergence condition, the convergence condition comprising that a number of times of iterative updating reaches a threshold number of times, or a change rate of the second probability distribution obtained by two adjacent times of iterative updating is less than a threshold change rate.

[0029] Optionally, the first determination unit is further configured to:

[0030] determine the relationship type corresponding to a maximum probability value in the second probability distribution as the second relationship type.

[0031] Optionally, the apparatus further comprises:

[0032] a fifth determination module, configured to determine the second relationship type between the second related account and the target account as a default relationship type in response to the second related account not having the historical interaction record with the first related account.

[0033] Optionally, the first determining module comprises:

[0034] The second determining unit is configured to determine an account in the account relationship network as the associated account corresponding to the target account, if the account has the historical interaction record with the target account.

[0035] The third determining unit is configured to determine the associated account as the primary associated account, if the number of interactions between the associated account and the target account is greater than a threshold number, and the historical interaction record contains preset interaction information, wherein the preset interaction information is generated from the interaction between the accounts or is obtained by collecting the interaction content, the account information of the interaction participants and the device information from the interaction process.

[0036] The fourth determining unit is configured to determine the associated account as the secondary associated account, if the number of interactions between the associated account and the target account is less than the threshold number, or the historical interaction record does not contain the preset interaction information.

[0037] Optionally, the preset interaction information comprises at least one of the following: interaction mode, interaction message, the historical login device of the interaction initiator and the interaction receiver, the account type of the interaction receiver, and the historical WiFi connection of the interaction initiator and the interaction receiver.

[0038] Optionally, the second determining module comprises:

[0039] The classification unit is configured to input the preset interaction information between the primary associated account and the target account into a relationship classification model to obtain the primary relationship type output by the relationship classification model, wherein the relationship classification model is trained according to sample interaction information and sample relationship types.

[0040] The fifth determining unit is configured to determine the primary probability distribution according to the primary relationship type and the number of relationship types.

[0041] Optionally, the historical interaction record comprises at least one of the following: historical resource transfer record, historical resource exchange record, and historical resource sharing record.

[0042] In another aspect, an embodiment of the present application provides a server, comprising a processor and a memory, wherein the memory stores at least one instruction, at least one program, a code set or an instruction set, and the at least one instruction, the at least one program, the code set or the instruction set is loaded and executed by the processor to implement the account relationship determining method as described in the above aspect.

[0043] In another aspect, an embodiment of the present application provides a computer readable storage medium, which stores at least one instruction, at least one program, a code set or an instruction set, and the at least one instruction, the at least one program, the code set or the instruction set are loaded and executed by a processor to implement the account relationship determination method according to the above aspect.

[0044] According to an aspect of the present application, a computer program product or computer program is provided, which includes computer instructions stored in a computer readable storage medium. A processor of a server reads the computer instructions from the computer readable storage medium, and the processor executes the computer instructions to enable the server to perform the account relationship determination method provided in various optional implementations of the above aspect.

[0045] The technical scheme provided by the embodiments of the present application has at least the following beneficial effects:

[0046] In the embodiments of the present application, the relationship type corresponding to the first associated account with strong interaction with the target account is first expressed by a first probability distribution, and a second probability distribution is further determined based on the historical interaction records between the first associated account and the second associated account, so that the second relationship type between the second associated account and the target account is obtained according to the second probability distribution. The relationship type between other accounts is determined by using the probability distribution transmission mode from the account relationship with high interaction and easy to determine, without obtaining the detailed information of all accounts in the account relationship network, reducing the cost of account relationship determination, and obtaining the relationship type between each pair of interactive accounts in the account relationship network, improving the efficiency and accuracy of determining the account relationship. BRIEF DESCRIPTION OF DRAWINGS

[0047] Figure 1 is a schematic diagram of an implementation environment provided by an exemplary embodiment of the present application;

[0048] Figure 2 is a flowchart of an account relationship determination method provided by an exemplary embodiment of the present application;

[0049] Figure 3 is a schematic diagram of an account relationship network provided by an exemplary embodiment of the present application;

[0050] Figure 4 is a flowchart of an account relationship determination method provided by another exemplary embodiment of the present application;

[0051] Figure 5 is a schematic diagram of an ego-centric network provided by another exemplary embodiment of the present application;

[0052] Figure 6is a flow chart of the account relationship determination method provided by another exemplary embodiment of the present application;

[0053] Figure 7 is a flow chart of the account relationship determination method provided by another exemplary embodiment of the present application;

[0054] Figure 8 is a structural block diagram of the account relationship determination device provided by an exemplary embodiment of the present application;

[0055] Figure 9 is a structural block diagram of the server provided by an exemplary embodiment of the present application. DETAILED DESCRIPTION

[0056] In order to make the purpose, technical solutions and advantages of the present application more clear, the embodiments of the present application will be further described in detail below with reference to the drawings.

[0057] In the present text, “multiple” refers to two or more. “And / or” describes the association relationship of the associated objects, which means that there can be three relationships, for example, A and / or B can represent the three cases of A existing alone, A and B existing simultaneously, and B existing alone. The character “ / ” generally represents that the associated objects before and after it are in an “or” relationship.

[0058] The account relationship determination method provided by the embodiments of the present application can be applied to the determination of the relationship between network accounts. The following will be described in combination with several application scenarios.

[0059] 1. Account anomaly detection scenario

[0060] In the account anomaly detection scenario, the method provided by the embodiment of the present application can be applied to the background server of a fund transaction platform (such as an e-commerce transaction platform, a mobile payment platform, an online bank, etc.). The background server first acquires the historical interaction records of the abnormal account, and determines the first associated account and the second associated account of the abnormal account from the account relationship network, wherein the interaction between the first associated account and the target account is relatively strong, and the background server can directly determine the first relationship type between the first associated account and the abnormal account according to the historical interaction records. For the second associated account with weak interaction, the background server performs probability diffusion on the first probability distribution corresponding to the first relationship type and the historical interaction records between the first associated account and the second associated account, determines the second relationship type between the second associated account and the target account according to the updated second probability distribution, and thus obtains the relationship type (such as including a commercial relationship and a non-commercial relationship) between the abnormal account and each associated account, and determines the associated account with the relationship type between the abnormal account belonging to the commercial relationship and the non-commercial relationship. In a possible implementation manner, the background server determines the account relationship every predetermined time interval, so as to perform account anomaly detection based on the real-time updated account relationship network.

[0061] In the application stage, in the network fund transaction, when a certain account in the fund transaction platform has fund transaction anomaly, there is usually an account with fund transaction anomaly in the associated account that has fund transaction with the abnormal account, therefore, the background server screens out the account with the commercial relationship between the abnormal account based on the determination result of the account relationship, and feeds back to the risk detection mechanism of the fund transaction platform. The account relationship network is used to mine the abnormal account, and the efficiency of the account anomaly detection and the security of the fund transaction platform are improved.

[0062] 2. Financial product recommendation scenario

[0063] In the financial product recommendation scenario, the method provided by the embodiment of the present application can be applied to the background server of a fund transaction platform (such as an e-commerce transaction platform, a mobile payment platform, an online bank, etc.). In reality, the fund transactions between multiple accounts with the same type of social relationship are more close, therefore, in the process of gradually determining the relationship type between the accounts by performing probability distribution variable diffusion and iteration according to the relationship type of the first associated account, a relationship community is formed, such as a family community, a friend community, a work community, a commercial cooperation community, etc. The background server can determine the relationship type between the first associated account and the target account based on the features corresponding to each relationship type, and determine the relationship type between the second associated account and the target account through the diffusion of the probability distribution.

[0064] In the application stage, different types of financial products and consumption coupons are recommended to accounts with different relationship types. For example, for family financial management products or friend cooperation marketing activities, the background server recommends them to accounts with relationship types such as family relationship, work relationship, or friend relationship. For commercial financial products such as commercial consumption coupons, the background server can recommend them to accounts corresponding to business relationships.

[0065] 3. Model training scenario

[0066] In the model training scenario, the method provided by the embodiments of the present application can be applied to a computer device for training a neural network model (such as a marketing recommendation model or a financial default model). Taking the marketing recommendation model as an example, the computer device obtains account interaction data of different relationship types and inputs them as sample data into the marketing recommendation model to be trained. Using a large amount of account interaction data of different relationship types as sample data can improve the accuracy of the neural network model.

[0067] The above is only illustrative with several common application scenarios as examples. The method provided by the embodiments of the present application can also be applied to other scenarios that need to determine the relationship between accounts, and the embodiments of the present application do not limit the actual application scenarios.

[0068] In the related art, when the server detects the interaction behavior between accounts, it obtains detailed information of the two accounts, such as the login location of the account, the name, age, and home address of the user corresponding to the account, and determines the relationship type between the two based on the detailed information of the two. Or, the server directly extracts features from the interaction behavior between the target account and the associated account, and performs clustering, and determines that the associated accounts with high feature similarity belong to the same category of associated accounts, i.e., the same category of associated accounts and the target account belong to the same community.

[0069] However, if the relationship type is determined based on detailed account information, the relationship type between the target account and the account with incomplete information or low interaction frequency cannot be determined. If clustering is directly performed based on the interaction behavior features, the relationship between specific users cannot be determined, and the efficiency and accuracy of the relationship classification are low.

[0070] To solve the problems in the prior art, the account relationship determination method provided in the present application is provided, the background server of the network interaction platform records the interaction between each account in real time, when the instruction of account relationship determination is executed, first, the relationship type corresponding to the first associated account with strong interaction with the target account is expressed by the first probability distribution, and based on the historical interaction record between the first associated account and the second associated account, the first probability distribution is used for probability diffusion to further determine the second probability distribution, so that the second relationship type between the second associated account and the target account is obtained according to the second probability distribution, and the relationship type between other accounts is determined from the account relationship with high interaction and easy to determine, and different from the related art, the account relationship determination method provided in the present application does not need to obtain the detailed information of all accounts, and the relationship type between each pair of interactive accounts in the account relationship network can be obtained, the difficulty and implementation cost of account relationship determination are reduced, and the efficiency and accuracy of account relationship determination are improved.

[0071] Figure 1 A schematic diagram of an implementation environment provided by an embodiment of the present application is shown. The implementation environment can include: a first terminal 110, a server 120 and a second terminal 130.

[0072] The first terminal 110 is installed and runs a client 111 supporting interactive operation, when the first terminal 110 receives the interactive initiation operation or the interactive reception operation of the first user 112, the client 111 sends the interaction data to the server 120, and the server 120 records the interaction of the account corresponding to the user 112 according to the received interaction data.

[0073] The second terminal 130 is installed and runs a client 131 supporting interactive operation, when the second terminal 130 receives the interactive initiation operation or the interactive reception operation of the second user 132, the client 131 sends the interaction data to the server 120, and the server 120 records the interaction of the account corresponding to the user 132 according to the received interaction data.

[0074] Figure 1 Only two terminals are shown, and in other embodiments, more terminals can also be connected to the server 120.

[0075] The first terminal 110, the second terminal 130 and other terminals are connected to the server 120 through a wireless network or a wired network.

[0076] The server 120 comprises a memory 121, a processor 122, an account database 123, a relationship determining module 124, and a user-oriented input / output interface (I / O interface) 125. The processor 122 is configured to load instructions stored in the server 120 and process data in the account database 123 and the relationship determining module 124. The account database 123 is configured to store account information and historical interaction records of the first terminal 110, the second terminal 130, and other terminals, such as account identifiers, account types, interaction times, interaction contents, historical login devices corresponding to the accounts, and historical connection wireless fidelity (WiFi), etc. The relationship determining module 124 is configured to determine relationship types between a target account and primary associated accounts and secondary associated accounts according to historical interaction data stored in the account database 123. The user-oriented I / O interface 125 is configured to establish communication with the first terminal 110 and / or the second terminal 130 through a wireless network or a wired network and exchange data.

[0077] In an illustrative example, the server 120 obtains interaction records between accounts in real time, and when receiving an instruction to determine account relationships, the server 120 determines primary associated accounts and secondary associated accounts corresponding to a target account. For a primary associated account with high interaction with the target account, the server 120 directly determines a primary relationship type between the primary associated account and the target account according to preset interaction information, and determines a primary probability distribution corresponding to the primary associated account based on the primary relationship type. For a secondary associated account with weak interaction with the target account, the server 120 initializes a secondary probability distribution of the secondary associated account based on historical interaction records between the secondary associated account and the primary associated account and the primary probability distribution, and iteratively updates the secondary probability distribution multiple times according to historical interaction records between the primary associated account and the secondary associated account and historical interaction records between the secondary associated accounts, so as to obtain a secondary probability distribution close to a real situation, and determines a secondary relationship type between the secondary associated account and the target account according to the secondary probability distribution.

[0078] Figure 2 A flowchart of an account relationship determining method provided by an example embodiment of the present application is shown. In this embodiment, the method is used to determine relationships between accounts in an account relationship network. Figure 1 The server 120 in the implementation environment shown is taken as an example for illustration, and the method comprises the following steps.

[0079] In step 201, primary associated accounts and secondary associated accounts corresponding to a target account are determined from an account relationship network. The interaction between the primary associated accounts and the target account is higher than the interaction between the secondary associated accounts and the target account. The account relationship network is a relationship network formed based on historical interaction records between accounts.

[0080] The account relationship network represents the network of relationships formed by interactions between multiple accounts on the same online platform. Within this network, every account has a historical interaction record with at least one other account. The backend server records the account information of both parties during the interaction, determining the existence of a relationship between them, thus gradually forming the account relationship network. For a given account in the network, accounts with historical interaction records with it are designated as its neighboring accounts. This account and its neighboring accounts together form an ego-network centered on that account. In the ego-network, the central account has historical interaction records with all other accounts, while neighboring accounts may or may not have historical interaction records with each other.

[0081] Indicative, such as Figure 3 As shown, it illustrates a schematic diagram of an account relationship network. The interaction between multiple accounts forms this account relationship network. There is a historical interaction record between two accounts connected by a line segment. It can be seen that any account has a historical interaction record with at least one other account. Figure 3 A schematic diagram of an Ego Network centered on target account 301 is also shown. In this Ego Network, all accounts other than target account 301 have historical interaction records with target account 301.

[0082] In actual online interaction platforms, as the number of interactions between accounts increases and the network of account relationships expands, there are usually first-level related accounts with strong interaction and second-level related accounts with weak interaction among the related accounts that have historical interaction records with the target account. The interaction between the target account and the first-level related account is more frequent and the information in the historical interaction records is more complete.

[0083] In one possible implementation, the server determines the primary and secondary associated accounts based on the historical interaction records corresponding to the target account. For example, associated accounts that meet preset conditions are determined as primary associated accounts, and other accounts in EgoNetwork besides the primary associated accounts are determined as secondary associated accounts.

[0084] Step 202: Determine the first-level probability distribution of the first-level associated account based on the first-level relationship type between the first-level associated account and the target account. The first-level probability distribution is used to characterize the probability distribution of the relationship type between the first-level associated account and the target account. The relationship type includes at least two types.

[0085] In a possible implementation, the embodiments of the present application represent the relationship type between the associated account and the target account by a probability distribution, in which the sum of the probability values corresponding to each relationship type is 1. Since the first-level relationship type between the first-level associated account and the target account has been determined, in the first-level probability distribution, the probability of the relationship type belonging to the first-level relationship type is 1, and the probability of the relationship type belonging to other relationship types is 0.

[0086] Illustratively, if the preset relationship types only include two types, the first-level probability distribution adopts a binomial distribution; if the preset relationship types include at least three types, the first-level probability distribution adopts a multinomial distribution.

[0087] For example, in a mobile payment network platform, the relationship types between accounts include commercial relationship and non-commercial relationship, or are specifically divided into family relationship, colleague relationship, friend relationship and commercial relationship, etc. The developers pre-set the relationship types according to the actual needs of account classification, and determine the corresponding probability distribution.

[0088] When the relationship types include two types a and b, and the first-level relationship type belongs to the relationship type a, the expression of the first-level probability distribution is as follows:

[0089]

[0090] In the formula, Pr(X=a) represents the probability of the first-level relationship type belonging to the relationship type a, and Pr(X=b) represents the probability of the first-level relationship type belonging to the relationship type b.

[0091] When the relationship types include at least three relationship types a, b, c and other relationship types, and the first-level relationship type belongs to the relationship type a, the expression of the first-level probability distribution is as follows:

[0092]

[0093] In the formula, Pr(X=a) represents the probability of the first-level relationship type belonging to the relationship type a, Pr(X=b) represents the probability of the first-level relationship type belonging to the relationship type b, and Pr(X=c) represents the probability of the first-level relationship type belonging to the relationship type c.

[0094] In step 203, based on the first-level probability distribution and the historical interaction record between the first-level associated account and the second-level associated account, a second-level probability distribution of the second-level associated account is determined, which is used to represent the probability distribution of the relationship type between the second-level associated account and the target account.

[0095] Since the interaction between the secondary associated account and the target account is low, the server cannot directly determine the secondary relationship type, and in the account relationship network, the relationship type between the same community, i.e., the accounts with close interaction, is the same. Therefore, the server determines the secondary probability distribution by diffusing the primary probability distribution to the secondary associated account with historical interaction records between the primary associated account and the secondary associated account, to obtain the probability distribution of the relationship type between the secondary associated account and the target account.

[0096] As shown in FIG. 3, the account 302 is a primary associated account corresponding to the target account 301, and the secondary associated accounts corresponding to the target account 301 include the account 303, the account 304, the account 305, and the account 306. The account 303 and the account 304 have historical interaction records with the account 302, and therefore, the server determines the secondary probability distribution of the account 303 and the account 304 according to the primary probability distribution of the account 302. Figure 3

[0097] Step 204: determining the secondary relationship type between the secondary associated account and the target account according to the secondary probability distribution.

[0098] The preset relationship type corresponding to the primary associated account and the secondary associated account is the same, and therefore, the type of the secondary probability distribution is the same as the primary probability distribution, i.e., the rule followed by the secondary probability distribution is the same as the primary probability distribution. The server determines the secondary relationship type between the secondary associated account and the target account according to the secondary probability distribution based on the rule of the primary probability distribution and the primary relationship type corresponding to the primary associated account.

[0099] Since the calculation number of the secondary probability distribution is low, the contingency of the account relationship determination result is too high, which may lead to the secondary relationship type between the secondary associated account and the target account being inconsistent with the actual relationship type. Therefore, in one possible implementation, the server calculates the secondary probability distribution multiple times until the secondary probability distribution approaches the actual probability distribution result, and then determines the secondary relationship type according to the secondary probability distribution.

[0100] In summary, in the embodiments of the present application, the relationship type corresponding to the primary associated account with strong interaction with the target account is described by the primary probability distribution, and the secondary probability distribution is further determined based on the historical interaction records between the primary associated account and the secondary associated account, so as to obtain the secondary relationship type between the secondary associated account and the target account according to the secondary probability distribution. The relationship type between the accounts is determined by the accounts with high interaction and easy determination in the manner of probability distribution transmission, without the need to obtain the detailed information of all the accounts in the account relationship network, which reduces the cost of account relationship determination and improves the efficiency and accuracy of the account relationship determination.​

[0101] In the process of determining the secondary probability distribution based on the primary probability distribution, the server cannot make the secondary probability distribution consistent with the actual secondary relationship type through one calculation because the secondary associated account can exist historical interaction records between multiple primary associated accounts with different primary probability distributions, so the server needs to constantly correct the secondary probability distribution according to the historical interaction records between the primary associated account and the secondary associated account and the historical interaction records between the secondary associated accounts until the secondary probability distribution converges.

[0102] Figure 4 A flowchart of a method for determining an account relationship provided by another example embodiment of the application is shown. The embodiment takes the method for determining an account relationship in an account relationship network. Figure 1 The server 120 in the implementation environment shown is taken as an example for illustration, and the method includes the following steps:

[0103] Step 401, determining the primary associated account and the secondary associated account corresponding to the target account from the account relationship network, the interaction of the primary associated account with the target account is higher than the interaction of the secondary associated account with the target account, and the account relationship network is a relationship network formed based on historical interaction records between accounts.

[0104] Step 402, determining the primary probability distribution of the primary associated account according to the primary relationship type between the primary associated account and the target account, the primary probability distribution is used to represent the probability distribution of the relationship type between the primary associated account and the target account, and the relationship type includes at least two types.

[0105] The specific implementation of steps 401 to 402 can refer to steps 201 to 202 described above, and the embodiments of the application will not be described here.

[0106] Step 403, in response to the existence of historical interaction records between the secondary associated account and the primary associated account, initializing the secondary probability distribution according to the primary probability distribution.

[0107] In the process of probability diffusion using the primary probability distribution, the secondary probability distribution obtained by initialization is meaningful only when the primary associated account and the secondary associated account exist historical interaction records, so the server obtains the historical interaction information corresponding to each primary associated account, and initializes the secondary probability distribution for the secondary associated account with historical interaction records with the primary associated account.

[0108] The interaction between the secondary associated account and the target account is low, and the server cannot directly determine the secondary probability distribution, so the server needs to initialize the secondary probability distribution according to the primary probability distribution to obtain the secondary probability distribution after the initial probability transmission. In a possible implementation, step 403 includes the following steps:

[0109] Step 403a, determining a primary statistical variable corresponding to the primary associated account based on the primary probability distribution, different statistical variables corresponding to different relationship types.

[0110] Since the server needs to transmit the primary probability distribution to the secondary associated account to initialize the secondary probability distribution based on the relationship type of the primary associated account, if the complete primary probability distribution expression is transmitted to the secondary associated account, the calculation process is more complicated, and the primary probability distribution has the particularity that the probability value corresponding to the primary relationship type is 1 and the probability value corresponding to other relationship types is 0, therefore, the server sets a statistical variable for each associated account, different statistical variables are used to represent different relationship types, and the primary statistical variable corresponding to the primary associated account is obtained from the primary relationship type.

[0111] Illustratively, the relationship type includes two types, relationship type a and relationship type b, when the primary relationship type belongs to relationship type a, the server generates a primary statistical variable m=a, and when the primary relationship type belongs to relationship type b, the server generates a primary statistical variable m=b.

[0112] Step 403b, transmitting the primary statistical variable to the secondary associated account.

[0113] In a possible implementation, the server transmits the primary statistical variable to the secondary associated account which has a historical interaction record with the primary associated account. Illustratively, as shown in FIG. 3, the primary relationship type corresponding to the primary associated account 302 is relationship type a, the server generates a primary statistical variable m=a corresponding to the primary associated account 302, and transmits m=a to the account 303 and the account 304. Figure 3

[0114] Step 403c, initializing the secondary probability distribution according to the primary statistical variable received by the secondary associated account, and the probability of the relationship type in the secondary probability distribution is the ratio of the number of the primary statistical variable corresponding to different relationship types in the total number of the primary statistical variable.

[0115] After the transmission of the primary statistical variable is completed, the server needs to initialize the secondary probability distribution according to the primary statistical variable received by the secondary associated account. In a possible implementation, the server respectively represents the probability of each relationship type as the ratio of the number of the primary statistical variable corresponding to each relationship type in the total number of the primary statistical variable in the primary statistical variable received by the secondary associated account.​

[0116] For example, if the relationship type includes two, relationship type a and relationship type b, the expression of the secondary probability distribution is as follows:

[0117]

[0118] wherein, represents the number of primary statistical variables of m=a, represents the number of primary statistical variables of m=b.

[0119] If the relationship type includes at least three, relationship type a, relationship type b, relationship type c and other relationship types, the expression of the secondary probability distribution is as follows:

[0120]

[0121] wherein, represents the number of primary statistical variables of m=a, represents the number of primary statistical variables of m=b, represents the number of primary statistical variables of m=c.

[0122] Step 404, based on the historical interaction records between the primary associated account and the secondary associated account, and the historical interaction records between the secondary associated accounts, iteratively updating the secondary probability distribution.

[0123] When there are historical interaction records between the secondary associated account and multiple primary associated accounts, and the primary relationship types corresponding to the primary associated accounts are different, the secondary associated account will receive different primary statistical variables, and if only one calculation of the secondary probability distribution is performed, it will lead to the secondary probability distribution not consistent with the actual situation, and the server gradually approaches the probability distribution corresponding to the actual relationship type by multiple iterative updates of the secondary probability distribution.

[0124] In one possible implementation, step 404 includes the following steps:

[0125] Step 404a, determining the primary statistical variables based on the primary probability distribution, and determining the secondary statistical variables corresponding to the secondary associated account based on the secondary probability distribution, the primary statistical variables and the secondary statistical variables corresponding to the same relationship type are the same.

[0126] In the actual account interaction process, for the same target account, the plurality of accounts with the same social relationship are closely connected, and usually have historical interaction records. Therefore, in the Ego Network, the historical interaction records between the secondary associated accounts also affect the secondary probability distribution. In the iterative updating process of the secondary probability distribution, the server needs to combine the first statistical variable and the second statistical variable corresponding to the secondary associated account to re-determine the secondary probability distribution.

[0127] In a possible implementation, the server determines the second statistical variable for each secondary associated account that has been initialized with the secondary probability distribution according to the current secondary probability distribution. Different statistical variables correspond to different relationship types, and for the same relationship type, the value of the corresponding first statistical variable and the second statistical variable is the same.

[0128] The server determines the second statistical variable according to the relationship type with the highest probability value in the secondary probability distribution.

[0129] Illustratively, the relationship types include two types, relationship type a and relationship type b. If the probability value corresponding to the relationship type a is higher than the probability value corresponding to the relationship type b in the secondary probability distribution corresponding to a secondary associated account, the server generates the second statistical variable m=a.

[0130] In step 404b, the first statistical variable is transmitted to the secondary associated account, and the second statistical variable is transmitted to other secondary associated accounts that have historical interaction records with the secondary associated account.

[0131] The interaction between the secondary associated account and other secondary associated accounts affects the secondary probability distribution. Therefore, for the secondary associated account that has the secondary probability distribution, the server obtains the historical interaction record corresponding thereto, obtains other secondary associated accounts that have historical interaction records with the secondary associated account, and performs the transmission of the statistical variable based on the historical interaction record.

[0132] Illustratively, as shown in Figure 5 If the relationship types include two types, relationship type a and relationship type b, the account 502 is the first associated account of the target account 501, and the remaining accounts are secondary associated accounts, then after the initialization of the secondary probability distribution, the account 503 and the account 504 both correspond to the secondary probability distribution. In the subsequent iterative updating process of the secondary probability distribution, the server generates the second statistical variable according to the current secondary probability distribution of the account 503, transmits it to the account 504, generates the second statistical variable according to the current secondary probability distribution of the account 504, transmits it to the account 503, and still transmits the first statistical variable corresponding to the first associated account 502 to the account 503 and the account 504.

[0133] Step 404c, updating the secondary probability distribution according to the primary statistical variable and the secondary statistical variable received by the secondary associated account.

[0134] After the transmission of the primary statistical variable and the secondary statistical variable is completed, the server updates the secondary probability distribution according to the number of the primary statistical variable and the secondary statistical variable received by the secondary associated account.

[0135] For example, if the relationship type includes two types, relationship type a and relationship type b, the expression of the secondary probability distribution after the kth iteration of the initial secondary probability distribution is as follows:

[0136]

[0137] wherein k represents the number of iterations of the initial secondary probability distribution, represents the number of statistical variables received by the secondary associated account with m=a in the kth iteration, represents the number of statistical variables received by the secondary associated account with m=b in the kth iteration.

[0138] If the relationship type includes at least three types, relationship type a, relationship type b, relationship type c and other relationship types, the expression of the secondary probability distribution after the kth iteration of the initial secondary probability distribution is as follows:

[0139]

[0140] wherein, represents the number of statistical variables received by the secondary associated account with m=a in the kth iteration, represents the number of statistical variables received by the secondary associated account with m=b in the kth iteration, represents the number of statistical variables received by the secondary associated account with m=c in the kth iteration.

[0141] For example, as shown in FIG. 5B, the account 503 receives the statistical variables of the account 502 and the account 504, and the server updates the secondary probability distribution of the account 503 based on the statistical variables of the account 502 and the account 504. Figure 5

[0142] Step 405, in response to the secondary probability distribution satisfying the convergence condition, determining the secondary relationship type between the secondary associated account and the target account according to the secondary probability distribution, the convergence condition including that the number of iteration updates reaches the number threshold, or the change rate of the secondary probability distribution obtained by adjacent two iteration updates is less than the change rate threshold.

[0143] ​With the increase of the iteration number of the secondary probability distribution, the probability of each relationship type in the secondary probability distribution gradually approaches the actual situation, and the developer sets the convergence condition in advance according to the demand. When the server detects that the secondary probability distribution of each secondary associated account in the account relationship network meets the convergence condition, the secondary relationship type between the secondary associated account and the target account is determined according to the secondary probability distribution.

[0144] Illustratively, the convergence condition is that the iteration update number reaches 500 times, or the difference between the probability values corresponding to each relationship type in the secondary probability distribution obtained by the adjacent two iteration updates is less than 0.1.

[0145] In a possible implementation, step 405 includes the following steps:

[0146] Step 405a, determining the relationship type corresponding to the maximum probability value in the secondary probability distribution as the secondary relationship type.

[0147] The relationship type corresponding to the maximum probability value in the secondary probability distribution is the relationship type that the secondary relationship type is most likely to correspond to, so the server determines the relationship type corresponding to the maximum probability value in the secondary probability distribution as the secondary relationship type. For example, the secondary probability distribution of a certain secondary associated account is as follows:

[0148]

[0149] Then, it is determined that the relationship type between the secondary associated account and the target account is relationship type a.

[0150] Step 406, in response to the fact that there is no historical interaction record between the secondary associated account and the primary associated account, determining that the secondary relationship type between the secondary associated account and the target account is a default relationship type.

[0151] In the account relationship network, there may be a case that the secondary associated account only has a historical interaction record with the target account. At this time, the server cannot determine the secondary relationship type in the manner of steps 403 to 405. In a possible implementation, the server stores a default relationship type, which is determined by the developer in advance according to the relationship type that the secondary relationship type may correspond to when the secondary associated account only has a historical interaction record with the target account in the actual interaction process. When the server determines that there is no historical interaction record between the secondary associated account and the primary associated account, it is determined that the secondary associated account and the corresponding secondary relationship type are the default relationship type.

[0152] For example, in mobile payment platforms, the relationship between accounts is divided into commercial relationship and non-commercial relationship. When a secondary associated account only has historical interaction records with the target account, the secondary relationship between the secondary associated account and the target account is usually a commercial relationship. Therefore, when there are no historical interaction records between a secondary associated account and a primary associated account, the server determines that the secondary relationship between the secondary associated account and the target account is a commercial relationship.

[0153] Step 406 is parallel to steps 403 to 405 above. For different secondary associated accounts, the server executes steps 403 to 405, or executes step 406 to determine the secondary relationship type between the secondary associated account and the target account.

[0154] In this embodiment, when there are historical interaction records between secondary associated accounts and primary associated accounts, a primary statistical variable is first generated using the primary probability distribution corresponding to the primary associated account. The secondary distribution probability is initialized based on the historical interaction records between the primary associated account and the secondary associated account. This allows for probability diffusion of relationship pairs with strong interactivity to relationship pairs with weak interactivity and whose relationship types are not easily determined directly, without needing to obtain detailed information about each account to obtain its probability distribution. Furthermore, the secondary probability distribution is iteratively updated using the historical interaction records between the primary associated account and the secondary associated account, as well as between the secondary associated account and other secondary associated accounts, so that the secondary probability distribution gradually approximates the actual relationship type distribution, thereby improving the accuracy of account relationship determination.

[0155] exist Figure 2 On the basis of, such as Figure 6 The diagram illustrates a flowchart of an account relationship determination method provided in another exemplary embodiment of this application. This embodiment uses this method for... Figure 1 Taking server 120 in the implementation environment shown as an example, step 201 above includes steps 201a to 201c, and step 202 above includes steps 202a to 202b:

[0156] Step 201a: In the account relationship network, accounts that have historical interaction records with the target account are identified as associated accounts corresponding to the target account.

[0157] Among them, historical interaction records include at least one of historical resource transfer records, historical resource exchange records, and historical resource sharing records.

[0158] For an operation of transferring resources from an account to another account, such as a red packet sending or receiving operation, a fund transfer operation between accounts, and a virtual item gifting operation, the server generates a historical resource transfer record; for a resource exchange operation between the two accounts, such as a commodity purchase operation of an account in an e-commerce platform, or a payment operation of a user for a commodity purchased offline through a network account, the server generates a historical resource exchange record; for a resource sharing operation between multiple accounts, such as an operation of account A opening a kinship card for accounts B, C, and D, and an interactive operation of multiple accounts participating in a network activity to split rewards, the server generates a historical resource sharing record.

[0159] In step 201b, in response to the number of interactions between the associated account and the target account being greater than the number threshold, and the historical interaction record containing the preset interaction information, the associated account is determined as a first-level associated account.

[0160] The preset interaction information is generated by the interaction between the accounts, or is obtained by collecting the interaction content, account information, and device information of the interaction participants from the interaction process, and includes at least one of the interaction mode, the interaction message, the historical login device of the interaction initiator and the interaction receiver, the account type of the interaction receiver, and the historical connection WiFi of the interaction initiator and the interaction receiver.

[0161] The interaction mode is generated by the interaction behavior between the accounts. For example, for an interaction behavior in an instant messaging platform with payment function, when the interaction mode belongs to a transfer or red packet sending and receiving between contacts, the two parties usually belong to a non-commercial relationship; when the interaction mode belongs to a payment to a merchant, the two parties usually belong to a commercial relationship.

[0162] The interaction message is obtained by the server collecting the interaction content from the interaction process. When it is also a transfer behavior between accounts, the server can determine the relationship type between the two parties based on the message content, such as containing a call for relatives, friends, etc. in the message, the two parties usually belong to a non-commercial relationship, and when the interaction message contains keywords such as item name or price, cost, etc., the two parties usually belong to a commercial relationship.

[0163] The historical login device of the interaction initiator and the interaction receiver, the account type of the interaction receiver, and the historical connection WiFi of the interaction initiator and the interaction receiver are obtained by the server collecting the account information and device information of the interaction participants from the interaction process. For the interaction receiver and the interaction initiator with the same historical login device or a high similarity of the historical connection WiFi, they usually belong to a kinship relationship or a friendship relationship, etc. When the account type of the interaction initiator or the interaction receiver belongs to an enterprise account, they usually belong to a commercial relationship.

[0164] Illustratively, the historical interaction record also includes the account identifier of the interaction initiator, the account identifier of the interaction receiver, the interaction time, and other interaction content in addition to the interaction message, such as the transaction amount for a transfer or payment.

[0165] The server determines an associated account as a first-level associated account if the number of interactions between the associated account and the target account is greater than the number threshold and the associated account contains at least one of the preset interaction information.

[0166] In step 201c, the server determines the associated account as a second-level associated account in response to the number of interactions between the associated account and the target account being less than the number threshold or the historical interaction record not containing the preset interaction information.

[0167] When the historical interaction record between the associated account and the target account does not contain the preset interaction information, the server cannot determine the relationship type between the associated account and the target account according to the preset interaction information, and directly determines the associated account as a second-level associated account. If the number of interactions between the associated account and the target account is less than the number threshold, even if the historical interaction record contains the preset interaction information, the reference value of the preset interaction information is low when the number of interactions is small, for example, the historical interaction information contains less information about the historical connection of WiFi, resulting in the similarity calculation result of the historical connection of WiFi being accidental. Therefore, in order to avoid the server obtaining an incorrect relationship determination result according to less preset interaction information, the server determines the associated account with which the number of interactions with the target account is less than the number threshold as a second-level associated account.

[0168] In step 202a, the server inputs the preset interaction information between the first-level associated account and the target account into the relationship classification model to obtain a first-level relationship type output by the relationship classification model.

[0169] The relationship classification model is trained according to the sample interaction information and the sample relationship type.

[0170] In a possible implementation, the computer device used to train the relationship classification model is preconfigured with a corresponding relationship between the sample interaction information and the sample relationship type. For example, if the relationship type between the accounts is divided into a business relationship and a non-business relationship, the developer sets the sample relationship type corresponding to the sample interaction information in which the interaction participant account type is an enterprise account as a business relationship according to the account interaction characteristics corresponding to the two types of relationships; sets the sample relationship type corresponding to the sample interaction information in which there are the same historical login device or the similarity of the historical connection of WiFi is higher than a similarity threshold as a non-business relationship; and sets the sample relationship type corresponding to the sample interaction information in which the interaction mode is a business transaction (such as payment or purchase of goods) as a business relationship.

[0171] Illustratively, the relationship classification model is trained by the computer device using sample interaction information and sample relationship types, wherein the machine learning model is a model having a classification function, such as a convolutional neural network (CNN), a support vector machine (SVM), etc.

[0172] In step 202b, the primary probability distribution is determined according to the primary relationship type and the number of relationship types.

[0173] In one possible implementation, the primary probability distribution in the embodiments of the present application adopts a binomial distribution or a multinomial distribution, and the specific probability distribution function thereof needs to be determined by the server according to the number of preset relationship types. The server generates a corresponding primary probability distribution for each primary associated account according to the primary relationship type and the probability distribution function.

[0174] Illustratively, when the preset relationship types include two types, the probability distribution is determined to be a binomial distribution, and the server determines the primary probability distribution according to the primary relationship type and the binomial distribution function. When the preset relationship types include three or more types, the probability distribution is determined to be a multinomial distribution. In the primary probability distribution, the probability that a relationship type belongs to the primary relationship type is 1, and the probability that a relationship type belongs to other relationship types is 0.

[0175] In the embodiments of the present application, the server determines the primary associated account and the secondary associated account according to the number of interactions between the associated account and the target account and whether the preset interaction information is included, so as to avoid directly determining the relationship of the associated account which has a low interaction and a low reference of the preset interaction information, reduce the data processing pressure of the server, and ensure the accuracy of the relationship determination between the primary associated account and the target account. In addition, the server determines the primary relationship type between the primary associated account and the target account by using the pre-trained relationship classification model, thereby improving the efficiency and accuracy of the account relationship determination.

[0176] In combination with the above various embodiments, in one illustrative example, the flow of the account relationship determination is as shown in Figure 7

[0177] In step 701, the interaction records between accounts are obtained.

[0178] The server records the interaction process when detecting the interaction operation between accounts.

[0179] In step 702, an account relationship pair is generated.

[0180] The server determines the accounts having interaction records therebetween as an account relationship pair.​

[0181] Step 703, screening seed relationship pairs and determining relationship types of the seed relationship pairs.

[0182] In the seed relationship pairs, the two accounts are seed nodes to each other, and the seed node is the first-level associated account in the above embodiments.

[0183] Step 704, constructing an account relationship network.

[0184] Step 705, constructing a self-centered network for all accounts in the account relationship network.

[0185] Step 706, dividing all accounts into communities and updating a probability distribution of relationship types between the iteration non-seed relationship pairs.

[0186] In the non-seed relationship pairs, the two accounts are non-seed nodes to each other, and the non-seed node is the second-level associated account in the above embodiments. Dividing the accounts into communities means determining a first-level probability distribution of the seed relationship pairs according to the relationship types of the seed relationship pairs, initializing a second-level probability distribution of the non-seed relationship pairs according to the interaction records between the seed nodes and the non-seed nodes and the first-level probability distribution, and iteratively updating the second-level probability distribution according to the interaction records between the seed nodes and the non-seed nodes and the interaction records between the non-seed nodes.

[0187] Step 707, determining relationship types of all account relationship pairs.

[0188] According to the iteratively updated second-level probability distribution, the relationship types of the non-seed relationship pairs are determined, and thus the relationship types of all account relationship pairs are obtained.

[0189] Figure 8 is a structural block diagram of an account relationship determination device provided by an exemplary embodiment of the present application. The device comprises:

[0190] A first determination module 801 is configured to determine a first-level associated account and a second-level associated account corresponding to a target account from an account relationship network, the interaction between the first-level associated account and the target account is higher than the interaction between the second-level associated account and the target account, and the account relationship network is a relationship network formed based on historical interaction records between accounts.

[0191] A second determination module 802 is configured to determine a first-level probability distribution of the first-level associated account according to a first-level relationship type between the first-level associated account and the target account, the first-level probability distribution is used to represent a probability distribution situation of the relationship type between the first-level associated account and the target account, and the relationship type includes at least two types.

[0192] The third determining module 803 is configured to determine a second probability distribution of the second associated account based on the first probability distribution and the historical interaction record between the first associated account and the second associated account, where the second probability distribution is used to represent a probability distribution of a relationship type between the second associated account and the target account.

[0193] The fourth determining module 804 is configured to determine a second relationship type between the second associated account and the target account according to the second probability distribution.

[0194] Optionally, the third determining module 803 comprises:

[0195] An initializing unit, configured to initialize the second probability distribution according to the first probability distribution in response to the historical interaction record between the second associated account and the first associated account.

[0196] An updating unit, configured to iteratively update the second probability distribution based on the historical interaction record between the first associated account and the second associated account and the historical interaction record between the second associated accounts.

[0197] Optionally, the initializing unit is further configured to:

[0198] determine a first statistical variable corresponding to the first associated account based on the first probability distribution, where different statistical variables correspond to different relationship types;

[0199] transmit the first statistical variable to the second associated account;

[0200] initialize the second probability distribution according to the first statistical variable received by the second associated account, where the probability of a relationship type in the second probability distribution is a ratio of a number of the first statistical variable corresponding to the relationship type to a total number of the first statistical variables.

[0201] Optionally, the updating unit is further configured to:

[0202] determine the first statistical variable based on the first probability distribution and determine a second statistical variable corresponding to the second associated account based on the second probability distribution, where the first statistical variable and the second statistical variable corresponding to the same relationship type are the same;

[0203] transmit the first statistical variable to the second associated account and transmit the second statistical variable to other second associated accounts having the historical interaction record with the second associated account;

[0204] update the second probability distribution according to the first statistical variable and the second statistical variable received by the second related account.

[0205] Optionally, the fourth determining module 804 comprises:

[0206] The first determining unit is configured to, in response to the second probability distribution satisfying a convergence condition, determine the second relationship type between the second related account and the target account according to the second probability distribution, the convergence condition comprising that the number of times of iterative updating reaches a number threshold, or the change rate of the second probability distribution obtained by two adjacent times of iterative updating is less than a change rate threshold.

[0207] Optionally, the first determining unit is further configured to:

[0208] determine the relationship type corresponding to the maximum probability value in the second probability distribution as the second relationship type.

[0209] Optionally, the apparatus further comprises:

[0210] The fifth determining module is configured to, in response to the second related account and the first related account not existing the historical interaction record, determine the second relationship type between the second related account and the target account as a default relationship type.

[0211] Optionally, the first determining module 801 comprises:

[0212] The second determining unit is configured to determine an account in the account relationship network and existing the historical interaction record with the target account as a related account corresponding to the target account.

[0213] The third determining unit is configured to, in response to the number of interactions between the related account and the target account being greater than a number threshold, and the historical interaction record containing preset interaction information, determine the related account as the first related account, the preset interaction information being generated by interaction between accounts, or being obtained by collecting interaction content, account information and device information of an interaction participant in an interaction process;

[0214] The fourth determining unit is configured to, in response to the number of interactions between the related account and the target account being less than the number threshold, or the historical interaction record not containing the preset interaction information, determine the related account as the second related account.

[0215] Optionally, the preset interaction information comprises at least one of an interaction mode, an interaction message, a historical login device of an interaction initiator and an interaction receiver, an account type of the interaction receiver, and a historical connection wireless fidelity (WiFi) of the interaction initiator and the interaction receiver.

[0216] Optionally, the second determining module 802 comprises:

[0217] a classification unit, configured to input the preset interaction information between the primary associated account and the target account into a relationship classification model, to obtain the primary relationship type output by the relationship classification model, the relationship classification model being trained according to sample interaction information and a sample relationship type;

[0218] a fifth determining unit, configured to determine the primary probability distribution according to the primary relationship type and the number of relationship types.

[0219] Optionally, the historical interaction record comprises at least one of a historical resource transfer record, a historical resource exchange record and a historical resource sharing record.

[0220] To sum up, in the embodiments of the present application, the relationship type corresponding to the primary associated account with high interaction with the target account is expressed by the primary probability distribution, and the secondary probability distribution is further determined based on the historical interaction record between the primary associated account and the secondary associated account, so as to obtain the secondary relationship type between the secondary associated account and the target account according to the secondary probability distribution. The relationship type between other accounts is determined by the account relationship with high interaction and easy determination in a manner of probability distribution transmission, without the need to obtain detailed information of all accounts in the account relationship network, thereby reducing the cost of account relationship determination, and improving the efficiency and accuracy of account relationship determination.

[0221] Please refer to Figure 9 which shows a structural schematic diagram of a server provided by an embodiment of the present application. Specifically,

[0222] The server 900 comprises a central processing unit (CPU) 901, a system memory 904 comprising a random access memory (RAM) 902 and a read-only memory (ROM) 903, and a system bus 905 connecting the system memory 904 and the central processing unit 901. The server 900 further comprises a basic input / output (I / O) controller 906 helping to transfer information between various devices in the computer, and a mass storage device 907 for storing an operating system 913, application programs 914 and other program modules 915.

[0223] The basic input / output system 906 includes the various components that are used to display information, such as a display 908 and input devices 909, such as a mouse, keyboard, or electronic stylus, for inputting information. The display 908 and input devices 909 are connected to the central processing unit 901 through an input / output controller 910 that is connected to the system bus 905. The basic input / output system 906 can also include the input / output controller 910 for receiving and processing input from a number of other devices, including a keyboard, mouse, or electronic stylus. Similarly, the input / output controller 910 provides output to the display screen, a printer, or other type of output device.

[0224] The mass storage device 907 is connected to the central processing unit 901 through a mass storage controller (not shown) that is connected to the system bus 905. The mass storage device 907 and its associated computer readable media provide non-volatile storage for the server 900. That is, the mass storage device 907 can include a computer readable medium (not shown) such as a hard disk or a Compact Disc Read-Only Memory (CD-ROM) drive.

[0225] Without loss of generality, the computer readable media can include computer storage media and communication media. Computer storage media includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storage of information such as computer readable instructions, data structures, program modules or other data. Computer storage media includes RAM, ROM, Erasable Programmable Read Only Memory (EPROM), flash memory or other solid state memory technology, CD-ROM, Digital Video Disc (DVD), or other optical storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices. It should be understood by those skilled in the art that computer storage media does not limit the computer readable media to the foregoing examples. The system memory 904 and mass storage device 907 described above can be collectively referred to as memory.

[0226] According to various embodiments of the present application, the server 900 can also operate in a networking environment using logical connections to one or more remote computers, such as a host computer. The server 900 can connect to the network 912 through a network interface unit 911 connected to the system bus 905, which also allows the server 900 to be connected to other types of networks or remote computers (not shown).

[0227] The memory also includes at least one instruction, at least one program, a code set or an instruction set stored in the memory and configured to be executed by one or more processors to implement the account relationship determination method described above.

[0228] The embodiments of the present application also provide a computer readable storage medium, which stores at least one instruction, and the at least one instruction is loaded and executed by a processor to implement the account relationship determination method described in the above various embodiments.

[0229] According to an aspect of the present application, a computer program product or computer program is provided, which includes computer instructions stored in a computer readable storage medium. A processor of a server reads the computer instructions from the computer readable storage medium, and the processor executes the computer instructions, so that the server performs the account relationship determination method provided in various optional implementation manners of the above aspect.

[0230] Those skilled in the art should be aware that, in the above one or more examples, the functions described in the embodiments of the present application can be implemented in hardware, software, firmware or any combination thereof. When implemented in software, the functions can be stored in a computer readable storage medium or transmitted as one or more instructions or codes on a computer readable storage medium. The computer readable storage medium includes a computer storage medium and a communication medium, and the communication medium includes any medium that facilitates the transfer of computer programs from one place to another. The storage medium can be any available medium accessible by a general or special purpose computer.

[0231] The above description is only optional embodiments of the present application, and is not intended to limit the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method for determining account relationships, characterized in that, The method includes: The first-level and second-level associated accounts corresponding to the target account are determined from the account relationship network. The interaction between the first-level associated account and the target account is higher than that between the second-level associated account and the target account. The account relationship network is a relationship network formed based on the historical interaction records between accounts. The first-level associated account and the second-level associated account belong to an egocentric network centered on the target account. The central account in the egocentric network has historical interaction records with other accounts. Based on the first-level relationship type between the first-level associated account and the target account, a first-level probability distribution of the first-level associated account is determined. The first-level probability distribution is used to characterize the probability distribution of the relationship type between the first-level associated account and the target account. The relationship type includes at least two types. In response to the existence of historical interaction records between the secondary associated account and the primary associated account, a primary statistical variable corresponding to the primary associated account is determined based on the primary probability distribution, with different statistical variables corresponding to different relationship types; the primary statistical variable is transmitted to the secondary associated account; a secondary probability distribution is initialized based on the primary statistical variable received by the secondary associated account, wherein the probability of a relationship type in the secondary probability distribution is the ratio of the number of primary statistical variables corresponding to different relationship types to the total number of primary statistical variables, and the secondary probability distribution is used to characterize the probability distribution of the relationship types between the secondary associated account and the target account; In each round of iterative update, a primary statistical variable is determined based on the primary probability distribution, and a secondary statistical variable corresponding to the secondary associated account is determined based on the secondary probability distribution obtained in the previous round of iterative update. The primary statistical variable and the secondary statistical variable corresponding to the same relationship type are the same. The primary statistical variable is passed to the secondary associated account, and the secondary statistical variable is passed to other secondary associated accounts that have the historical interaction record with the secondary associated account; The secondary probability distribution is updated based on the primary and secondary statistical variables received by the secondary associated account; The secondary relationship type between the secondary associated account and the target account is determined based on the secondary probability distribution.

2. The method according to claim 1, characterized in that, Determining the secondary relationship type between the secondary associated account and the target account based on the secondary probability distribution includes: In response to the convergence condition of the secondary probability distribution, the secondary relationship type between the secondary associated account and the target account is determined according to the secondary probability distribution. The convergence condition includes the number of iteration updates reaching a threshold, or the rate of change of the secondary probability distribution obtained from two adjacent iteration updates being less than a rate of change threshold.

3. The method according to claim 2, characterized in that, Determining the secondary relationship type between the secondary associated account and the target account based on the secondary probability distribution includes: The relation type corresponding to the maximum probability value in the second-level probability distribution is determined as the second-level relation type.

4. The method according to claim 1, characterized in that, The method further includes: In response to the absence of historical interaction records between the secondary associated account and the primary associated account, the secondary relationship type between the secondary associated account and the target account is determined to be the default relationship type.

5. The method according to claim 1, characterized in that, The step of determining the primary and secondary associated accounts corresponding to the target account from the account relationship network includes: In the account relationship network, accounts that have historical interaction records with the target account are identified as associated accounts corresponding to the target account; In response to the fact that the number of interactions between the associated account and the target account exceeds a threshold, and the historical interaction record contains preset interaction information, the associated account is determined as the first-level associated account. The preset interaction information is generated by the interaction between accounts, or is obtained by collecting interaction content, account information and device information of the interaction participants during the interaction process. In response to the fact that the number of interactions between the associated account and the target account is less than the number threshold, or that the preset interaction information is not included in the historical interaction records, the associated account is determined as the secondary associated account.

6. The method according to claim 5, characterized in that, The preset interactive information includes at least one of the following: interactive method, interactive message, historical login devices of the interactive initiator and the interactive receiver, account type of the interactive receiver, and historical Wi-Fi connections of the interactive initiator and the interactive receiver.

7. The method according to claim 1, characterized in that, The step of determining the first-level probability distribution of the first-level associated account based on the first-level relationship type between the first-level associated account and the target account includes: The preset interaction information between the primary associated account and the target account is input into the relationship classification model to obtain the primary relationship type output by the relationship classification model. The relationship classification model is trained based on the sample interaction information and the sample relationship type. The first-level probability distribution is determined based on the first-level relation type and the number of the relation types.

8. The method according to claim 1, characterized in that, The historical interaction records include at least one of the following: historical resource transfer records, historical resource exchange records, and historical resource sharing records.

9. An account relationship determination device, characterized in that, The device includes: The first determining module is used to determine the first-level associated account and the second-level associated account corresponding to the target account from the account relationship network. The interaction between the first-level associated account and the target account is higher than the interaction between the second-level associated account and the target account. The account relationship network is a relationship network formed based on the historical interaction records between accounts. The first-level associated account and the second-level associated account belong to an egocentric network centered on the target account. The central account in the egocentric network has historical interaction records with other accounts. The second determining module is used to determine the first-level probability distribution of the first-level associated account based on the first-level relationship type between the first-level associated account and the target account. The first-level probability distribution is used to characterize the probability distribution of the relationship type between the first-level associated account and the target account. The relationship type includes at least two types. The third determining module is used to respond to the existence of historical interaction records between the secondary associated account and the primary associated account, determine the primary statistical variable corresponding to the primary associated account based on the primary probability distribution, and different statistical variables correspond to different relationship types; transmit the primary statistical variable to the secondary associated account; initialize the secondary probability distribution according to the primary statistical variable received by the secondary associated account, wherein the probability of the relationship type in the secondary probability distribution is the ratio of the number of primary statistical variables corresponding to different relationship types to the total number of primary statistical variables, and the secondary probability distribution is used to characterize the probability distribution of the relationship types between the secondary associated account and the target account; In each round of iterative update, a primary statistical variable is determined based on the primary probability distribution, and a secondary statistical variable corresponding to the secondary associated account is determined based on the secondary probability distribution obtained in the previous round of iterative update. The primary statistical variable and the secondary statistical variable corresponding to the same relationship type are the same. The primary statistical variable is passed to the secondary associated account, and the secondary statistical variable is passed to other secondary associated accounts that have the historical interaction record with the secondary associated account; The secondary probability distribution is updated based on the primary and secondary statistical variables received by the secondary associated account; The fourth determining module is used to determine the secondary relationship type between the secondary associated account and the target account based on the secondary probability distribution.

10. A server, characterized in that, The server includes a processor and a memory, the memory storing at least one instruction, at least one program, code set, or instruction set, the at least one instruction, the at least one program, the code set, or instruction set being loaded and executed by the processor to implement the account relationship determination method as described in any one of claims 1 to 8.

11. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores at least one instruction, at least one program, code set, or instruction set, wherein the at least one instruction, the at least one program, the code set, or instruction set is loaded and executed by a processor to implement the account relationship determination method as described in any one of claims 1 to 8.

Images