WO2016029794A1 - 识别特征账号的方法及装置 - Google Patents

识别特征账号的方法及装置 Download PDF

Info

Publication number
WO2016029794A1
WO2016029794A1 PCT/CN2015/086617 CN2015086617W WO2016029794A1 WO 2016029794 A1 WO2016029794 A1 WO 2016029794A1 CN 2015086617 W CN2015086617 W CN 2015086617W WO 2016029794 A1 WO2016029794 A1 WO 2016029794A1
Authority
WO
WIPO (PCT)
Prior art keywords
node
account
tree
data
relationship network
Prior art date
Application number
PCT/CN2015/086617
Other languages
English (en)
French (fr)
Inventor
毛仁歆
何慧梅
王峰伟
何帝君
林瑞华
Original Assignee
阿里巴巴集团控股有限公司
毛仁歆
何慧梅
王峰伟
何帝君
林瑞华
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 阿里巴巴集团控股有限公司, 毛仁歆, 何慧梅, 王峰伟, 何帝君, 林瑞华 filed Critical 阿里巴巴集团控股有限公司
Publication of WO2016029794A1 publication Critical patent/WO2016029794A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q30/00—Commerce

Definitions

  • the present application relates to the field of Internet technologies, and in particular, to a method and apparatus for identifying a feature account.
  • the present application provides a new technical solution, which can solve the technical problem that the feature account cannot be accurately identified in the related art.
  • a method for identifying a feature account comprising:
  • the tree relationship network When the tree relationship network satisfies the preset feature condition, it is determined that the tree relationship network includes the feature account.
  • an apparatus for identifying a feature account comprising:
  • a network establishing unit mapping an association relationship between the transferee account corresponding to the first live transfer transaction record and the transferred party account as a correspondence between the parent node and the child node, and establishing a pair a tree-like relationship network;
  • the determining unit determines that the tree account network includes the feature account when the tree relationship network satisfies the preset feature condition.
  • the present application can accurately identify the feature account by analyzing the interaction data of the specific type of user account and based on the data flow link feature corresponding to the specific type of user account interaction data.
  • the application also converts the account relationship based on the specific type of user account interaction data into a tree-like relationship network, which facilitates the execution of relationship identification between accounts and helps improve the accuracy of identifying the feature account.
  • FIG. 1 shows a schematic flow chart of a method of identifying a feature account in accordance with an exemplary embodiment of the present invention
  • FIG. 2 shows a schematic diagram of a node relationship in accordance with an exemplary embodiment of the present invention
  • FIG. 3 shows a schematic flow chart of generating a tree-like relationship network by message passing according to an exemplary embodiment of the present invention
  • FIG. 4 illustrates a schematic diagram of initializing assignment of a node in accordance with an exemplary embodiment of the present invention
  • FIG. 5 illustrates a schematic diagram of generating a tree-like relationship network by message passing according to an exemplary embodiment of the present invention
  • FIG. 6 shows a schematic diagram of funds transfer in a tree-like relationship network, in accordance with an exemplary embodiment of the present invention
  • FIG. 7 is a schematic structural diagram of a server according to an exemplary embodiment of the present invention.
  • FIG. 8 shows a schematic block diagram of an apparatus for identifying a feature account in accordance with an exemplary embodiment of the present invention.
  • the present application determines the corresponding user by analyzing the interaction data of a specific type of user account.
  • the data flow chain formed between the accounts, and the data flow chain is generated into a tree-like relationship network, thereby identifying the feature account based on the characteristics of the tree-like relationship network.
  • FIG. 1 illustrates a method for identifying a feature account, which may be applied to a server, according to an exemplary embodiment of the present invention.
  • the method includes:
  • Step 102 Select a specific type of user account interaction data in the historical behavior data, and the interaction information from the data sender account received by the data receiver account corresponding to the specific type of user account interaction data is the data receiver account.
  • the specific type of user account interaction data reflects the data flow characteristics between the corresponding data receiver account and the data sender account. Based on the analysis of the data flow characteristics, it is determined whether the corresponding data interaction parties are Is a feature account.
  • the interaction data may be an instant messaging message transmitted between user accounts.
  • the interaction data may be a communication message transmitted between user accounts.
  • the interaction data may be a transfer transaction information between user accounts. .
  • Step 104 Mapping an association relationship between the data sender account corresponding to the specific type of user account interaction data and the data receiver account as a correspondence between the parent node and the child node, and establishing a corresponding tree shape. Relationship network
  • each data sender account can be "first live” through data interaction (because it is the first interactive message received by the data receiver account after registration, the message can be considered “first live” or “First activation” of the corresponding data receiver account) "multiple data receiver accounts, and each data recipient account that is first lived can also be used to first live other accounts, so the application can pass the "one" Correspondence and relationship transfer of multiple "one data sender account corresponding to multiple first-time data receiver accounts” (for example, the B account is first lived by the A account, and the B account can also live the C account first, thereby making " The first live relationship is transmitted between ABCs, and a corresponding tree-like relationship network is generated by a plurality of specific types of user account interaction data, and the tree-like relationship network It includes the data flow relationship when data interaction between each account is performed.
  • Step 106 When the tree relationship network satisfies a preset feature condition, determine that the tree relationship network includes a feature account.
  • the present application converts the first live relationship between each pair of accounts into a tree-like relationship network by analyzing the specific type of user account interaction data and utilizing the transfer characteristics of the first live relationship between the accounts.
  • the data flow feature included in the tree relationship network can accurately analyze whether the feature account is included in the tree relationship network.
  • the present application when constructing a tree relationship network, the present application may be implemented by using a parallel computing model, that is, improving the establishment efficiency of the tree relationship network by parallel computing.
  • a parallel computing model that is, improving the establishment efficiency of the tree relationship network by parallel computing.
  • the tree-relational network may be established by using a BSP (Bulk Synchronous Parallel Computing Model).
  • BSP Bit Synchronous Parallel Computing Model
  • FIG. 2 it is assumed that a specific type of user account interaction data in the historical behavior data is extracted, and the first live relationship between the accounts shown in FIG. 2 is obtained by analyzing the user account interaction data, such as FIG. 2 (a The node A first live node B, the node A first live node C, the node C first live node D shown in Fig. 2(b), and the node D first live node shown in Fig. 2(c) E. The node F is first lived by the node D.
  • the tree relationship between multiple nodes can be easily derived from Figure 2, based on a large number of specific types of user account interaction data, the number of first live relationships between the accounts obtained by the server is also very large, then the server only It is possible to know whether there is a first-live relationship between each pair of accounts, but it is not easy to derive the entire tree-like relationship network.
  • FIG. 3 is a schematic flowchart of generating a tree relationship network by message delivery according to an exemplary embodiment of the present invention, including:
  • Step 302 an initialization operation: all accounts corresponding to a specific type of user account interaction data
  • the number is mapped to the corresponding node, and each node is configured with a corresponding unique identifier and a unique value.
  • the corresponding unique identifiers are A, B, C, etc.; and the unique value of each node may be in the form of: self ID + "#" + superstep (super step) steps,
  • the ID is the unique identifier of each node, such as A, B, C, etc. above, superstep refers to the step of each iteration operation, and "#" is used to distinguish the values of "ID” and "superstep". It can be seen that when the number of steps of ID and superstep is easily distinguished, "#" can also be omitted.
  • the unique value corresponding to node A is A#0
  • the unique value corresponding to node B is B#0
  • the unique value corresponding to node C is C#0
  • the unique value corresponding to node D is E#0
  • the unique value corresponding to node F is F#0.
  • each node directly passes a message containing its own unique identifier to the downstream node.
  • the downstream node is a node directly connected to the local node by a one-way edge, as shown in FIG. 5: in FIG. 5 (a1), node A directly transfers "A" to node B and node C, and node B does not exist. Downstream node; in Figure 5 (b1), node C passes "C” directly to node D; in Figure 5 (c1), node D directly passes "D" to node E and node F, node E and node F does not have a downstream node.
  • steps 306 to 320 it is necessary to implement the delivery of the message on the node by using multiple iterations. Then, each iteration will perform step 306 to step 320, which will be described in detail below with reference to FIG. 5.
  • the iterative operation includes:
  • the node B and the node C receive the message “A” from the node A;
  • the node D receives the message "C” from the node C;
  • the node E and the node F receive the message "D" from the node D; meanwhile, the node A No message was received.
  • Step 308 for the node that received the message in step 306, update its own unique value to: ID value + "#" + current superstep in the received message.
  • Step 310 Determine whether the node that completes the update of its unique value in step 308 has a corresponding downstream node. If yes, go to step 312, otherwise go to step 314.
  • the node B does not have a downstream node
  • the node C has a downstream node as a node D
  • a node D has a downstream node as node E and node F
  • node E and node F do not have a downstream node.
  • Step 312 in step 310, it is determined that there is a node of the downstream node, and the message received by itself is continuously transmitted to the downstream node.
  • node C passes message "A” to node D; in FIG. 5 (c2), node D passes message "C” to node E and node. F.
  • step 314 for the node that does not receive the message in step 306, or the node that does not have the downstream node in step 310, the processing of the node is ended in the current superstep. Specifically, node A does not receive the message, and node B, node E, and node F do not have a downstream node.
  • step 316 it is determined whether there is no message sent between all the nodes, or whether the number of iterations of the superstep has reached the maximum number of iterations. If no message is sent or the maximum number of iterations is reached, then the process proceeds to step 318, otherwise, the process proceeds to step 320. .
  • Step 318 ending and outputting a unique value for each node.
  • the condition for proceeding to step 318 has not been met at this time.
  • the iterative operation includes:
  • node D receives the message "A" from node C;
  • node E and node F receive the message "C” from node D; meanwhile, node A, node B, and node C have not received the message.
  • Step 308' for the node that received the message in step 306', updates its own value to: ID value + "#" + current superstep in the received message.
  • Step 310' it is determined whether the node that completes its own unique value update in step 308' has a corresponding downstream node, if yes, then proceeds to step 312', otherwise proceeds to step 314'.
  • the node D has a downstream node as a node E and a node F, and the node E and the node F do not have a downstream node.
  • Step 312' in step 310', it is determined that there is a node of the downstream node, and its own value is passed to the downstream node.
  • the node D transfers "A" to the node E and the node F.
  • Step 314' for the node that did not receive the message in step 306', or the node that does not have the downstream node in step 310', ends the processing of the node in the current superstep. Specifically, node A, node B, and node C do not receive a message, and node E and node F do not have a downstream node.
  • Step 316 ′ determining whether there is no message transmission between all the nodes, or whether the number of iterations of the superstep has reached the maximum number of iterations. If no message is sent or the maximum number of iterations is reached, then the process proceeds to step 318 ′, otherwise, the process proceeds to Step 320'.
  • Step 318' ends and outputs a unique value for each node.
  • the condition for the transition to step 318' has not been met at this time.
  • the iterative operation includes:
  • Step 308 for the node that received the message in step 306", update its own unique value to: ID + "#” + current superstep in the received message.
  • Step 310 it is determined whether the node that completes the update of its own value in step 308" has a corresponding downstream node, if yes, then proceeds to step 312", otherwise proceeds to step 314".
  • step 314" the node E and the node F do not have a downstream node, and directly proceeds to step 314".
  • Step 314" for the node that does not receive the message in step 306", or the node that does not have the downstream node in step 310", ends the processing of the node in the current superstep. Specifically, node A, node B, node C And node D did not receive the message, and node E and node F did not have a downstream node.
  • Step 316 determining whether no messages are sent between all the nodes, or whether the number of iterations of the superstep has reached the maximum number of iterations. If no message is sent or the maximum number of iterations is reached, then the process proceeds to step 318 ”, otherwise, the process proceeds to Step 320". Since no messages are sent by all the nodes in this superstep, the process proceeds to step 318".
  • Step 318 ending and outputting a unique value for each node.
  • the unique value of the node A is A#0, and the node The unique value of B is A#1, the unique value of node C is A#1; in Figure 5(b4), the unique value of node D is A#2; in Figure 5 (c4), node E and node F The only value is A#3.
  • the server learns that there is a tree-like relational network with node A as the root node according to the unique value obtained by each node, and node B and node C are the first-level child nodes of node A, and node D is the second-level child of node A.
  • the node, the node E and the node F are three-level child nodes of the node A, and form a tree-like relationship network as shown in FIG. 5(d).
  • a tree-like relationship network can be obtained. For each tree-like relationship network, whether the feature account is included or not can be determined according to the characteristics of the network.
  • the determination may be made according to the number of nodes included in each tree-like relationship network. Specifically, since the relationship between nodes is relatively simple in a tree-like relationship network based on normal transactions, there is no case where many nodes are associated with each other at multiple levels, and thus each tree-like relationship network can be acquired. If the number of nodes is greater than or equal to the preset number threshold, it can be determined that the corresponding tree-like relationship network includes the node corresponding to the feature account.
  • the determination may be made according to the maximum tree growth rate of each tree-like relationship network.
  • each tree-like relationship network can be acquired within a unit time length.
  • the maximum node increase speed for example, the maximum single day (that is, the unit time length is daily) node increase number, then when the maximum node increase speed is greater than or equal to the preset speed threshold, it can be determined that the corresponding tree relationship network is included There are nodes corresponding to the feature account.
  • the present application can be applied to various types of data interaction scenarios.
  • the technical solutions of the present application are described in detail below by taking typical applications therein as an example.
  • a specific type of user account interaction data may be a transfer transaction information between user accounts.
  • “feature account” That is, there is a fake transaction account, such as the seller user himself or a third party to control a series of virtual buyer accounts, and through these virtual buyer accounts on the Alipay platform for false transactions, resulting in the seller's reputation is high, the product rankings rise It is not conducive to the buyer's user to make a correct judgment; in addition, a malicious user controls a series of virtual buyer accounts, and maliciously defrauds the seller's marketing resources through these virtual buyer accounts to illegally obtain improper profits.
  • the virtual buyer account involved in the above false transaction can be effectively identified, thereby making a reasonable adjustment to the weight of the search, purchase, and the like.
  • the amount of the transfer between the nodes in the tree-like relationship network may be used to determine whether the feature account is included.
  • the specific type of user account interaction data may be the first live transfer transaction record, that is, the transferee account corresponding to the first live transfer transaction record is activated for the first time by the transfer operation of the transferee account. .
  • the transfer amount corresponding to all the first live transfer transaction records involved is obtained, and the difference between the corresponding transfer amount is less than or equal to the preset difference threshold.
  • Transaction records are treated as the same group. Assuming that the preset difference threshold is "0.5 million", since the transfer amount between node a and node b is 200,000, the transfer amount between node b and node e is 198,000, and between node e and node i.
  • the transfer amount is 202,000, the transfer amount between node i and node k is 200,000, and the transfer amount between node k and node n is 198,000, then the corresponding five first live transfer transaction records are determined as a group;
  • the transfer amount between node b and node d is 30,000, and the transfer amount between node e and node h is 30,000, then the corresponding two first live transfer transaction records are determined as one group; meanwhile, due to node e and node
  • the transfer amount between j is 60,000, the transfer amount between node j and node l is 5.8, and the corresponding two first live transfer transaction records are determined as one group; and the first live transfer transaction records corresponding to other transfer amounts respectively form one Group, as shown in Table 1:
  • determining the tree relationship when there is a quantity of the first live transfer transaction record included in the at least one packet in a proportion of all the first live transfer transaction records being greater than or equal to a preset ratio threshold The feature account is included in the network. Assuming that the preset ratio threshold is 30%, since the proportion of the packet 1 is 41.7%>30%, it is determined that the tree structure network shown in FIG. 6 includes the feature account.
  • the account number of the transferred party corresponding to the first live transfer transaction record included in the packet whose "the proportion is greater than or equal to the preset ratio threshold" may be determined as a feature account, such as node b, node e in group 1.
  • the node i, the node k, and the node n can determine that the seller account corresponding to the node a performs a fake transaction through the account corresponding to the node, that is, the account is the virtual buyer account used by the seller account corresponding to the node a. Therefore, the preset permission of the feature account can be restricted; at the same time, the seller account corresponding to the node a can be processed, for example, the transaction completed by the feature account is not used to calculate the reputation, shipment amount and the like of the corresponding seller.
  • the nodes included therein may be used.
  • the number, the maximum node increase rate, or the proportion of the first live transfer transaction records of each group may also be combined with the above three means to determine whether the network account is included in the network; after determining the feature account in the tree relationship network, The feature account can be determined based on the grouping of the first live transfer transaction record.
  • the values of the preset number threshold, the preset difference threshold, the preset proportional threshold, and the preset speed threshold may be adjusted according to actual needs to control the strictness of selecting the feature account. Specifically, when the preset number threshold is larger, the preset difference threshold is larger, the preset ratio threshold is larger, and the preset speed threshold is larger, the corresponding selection criterion is stricter (ie, it is more difficult to determine that the feature account exists). The opposite is the looser (ie, it is easier to be identified as having a feature account).
  • the specific type of user account interaction data may be an instant messaging message or an account associated message between the user accounts.
  • the communication message includes an instant message and a non-immediate message, such as a message, a comment, and the like; the account associated message includes a friend add request, a group join request, a group member invitation request, an account attention operation, and the like.
  • a “feature account” is an account with fraud or harassment behavior, such as a user controlling a series of “small” and using these “small” to swindle or harass other users, such as sending fraudulent messages, Trojan websites. URL or ad, etc.
  • the account involved in the above fraud or harassment behavior can be effectively identified, thereby making a reasonable adjustment to the weight of the authority for sending a message, adding a friend, and the like.
  • the present application also proposes a schematic structural diagram of a server according to an exemplary embodiment of the present application shown in FIG. 7.
  • the server includes a processor, an internal bus, a network interface, a memory, and a non-volatile memory, and may of course include hardware required for other services.
  • the processor reads the corresponding computer program from the non-volatile memory into memory and then runs to form a device that identifies the feature account at a logical level.
  • the present application does not exclude other implementations, such as logic devices or software and hardware combinations, etc., that is, the execution body of the following processing flow is not limited. In each logical unit, it can also be hardware or logic device.
  • the device for identifying a feature account may include a record selection unit, a network establishment unit, and a determination unit. among them:
  • the record selection unit selects a specific type of user account interaction data in the historical behavior data, and the interaction information from the data sender account received by the data receiver account corresponding to the specific type of user account interaction data is the data receiver account.
  • the network establishing unit maps the association relationship between the data sender account corresponding to the specific type of user account interaction data and the data receiver account as a correspondence between the parent node and the child node, and establishes a corresponding tree.
  • the determining unit determines that the tree account network includes the feature account when the tree relationship network satisfies the preset feature condition.
  • the network establishing unit is specifically configured to:
  • each node when there is no parent node, each node sends a local delivery message to the corresponding child node; when the parent node exists, Each node receives a delivery message from the corresponding parent node, and takes the linear combination value of the unique identifier included in the received delivery message and the current number of iterations as the unique value of the corresponding node, and sends the received delivery message.
  • each node reports its unique value;
  • the node with the same unique identifier included in the reported unique value constitutes a corresponding tree-like relationship network, and the hierarchical state of each node in the tree-like relationship network corresponds to the current number of iterations included in the reported unique value.
  • the network establishing unit establishes the tree relationship network by using a parallel computing model.
  • the network establishing unit establishes the tree relationship network by using an overall synchronous parallel computing BSP model.
  • the preset feature condition includes:
  • the number of nodes included in the tree-like relationship network is greater than or equal to a preset number threshold.
  • the preset feature condition includes:
  • the maximum node increase speed of the tree-like relationship network is greater than or equal to a preset speed threshold.
  • the preset feature condition includes:
  • the specific type of user account interaction data is used as the same group.
  • There is a proportion of the specific type of user account interaction data contained in at least one packet in all of the specific types of user account interaction data is greater than or equal to a preset ratio threshold.
  • the determining unit is specifically configured to:
  • the present application determines the data flow chain formed between the corresponding user accounts by analyzing the interaction data of the specific type of user accounts, and generates the tree flow relationship network based on the tree relationship network.
  • the feature identifies the feature account.
  • a computing device includes one or more processors (CPUs), input/output interfaces, network interfaces, and memory.
  • processors CPUs
  • input/output interfaces network interfaces
  • memory volatile and non-volatile memory
  • the memory may include non-persistent memory, random access memory (RAM), and/or non-volatile memory in a computer readable medium, such as read only memory (ROM) or flash memory.
  • RAM random access memory
  • ROM read only memory
  • Memory is an example of a computer readable medium.
  • Computer readable media includes both permanent and non-persistent, removable and non-removable media.
  • Information storage can be implemented by any method or technology.
  • the information can be computer readable instructions, data structures, modules of programs, or other data.
  • Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read only memory. (ROM), electrically erasable programmable read only memory (EEPROM), flash memory or other memory technology, compact disk read only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, Magnetic tape cassette, magnetic tape storage or other magnetic A sexual storage device or any other non-transportable medium that can be used to store information that can be accessed by a computing device.
  • computer readable media does not include temporary storage of computer readable media, such as modulated data signals and carrier waves.

Landscapes

  • Business, Economics & Management (AREA)
  • Accounting & Taxation (AREA)
  • Development Economics (AREA)
  • Economics (AREA)
  • Finance (AREA)
  • Marketing (AREA)
  • Strategic Management (AREA)
  • Physics & Mathematics (AREA)
  • General Business, Economics & Management (AREA)
  • General Physics & Mathematics (AREA)
  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Information Transfer Between Computers (AREA)

Abstract

一种识别特征账号的方法及装置，包括：选取历史行为数据中的特定类型的用户账号交互数据（102），该特定类型的用户账号交互数据对应的数据接收方账号接收到的来自数据发送方账号的交互信息为所述数据接收方账号在注册后接收到的首条交互信息；将所述特定类型的用户账号交互数据对应的数据发送方账号与数据接收方账号之间的关联关系映射为父级节点与子级节点之间的对应关系，并建立对应的树状关系网络（104）；当所述树状关系网络满足预设特征条件时，判定该树状关系网络中包含特征账号（106）。在上述技术方案中，可以根据特定类型的用户账号交互数据对应的数据流动特征，自动识别出特征账号。

Description

识别特征账号的方法及装置 技术领域
本申请涉及互联网技术领域，尤其涉及识别特征账号的方法及装置。
背景技术
在互联网环境中，用户需要通过注册相应的账号，实现账号之间的数据交互。然而，部分用户通过注册或控制诸多账号，影响互联网的正常交互，但相关技术中无法准确地识别相应的特征账号。
发明内容
有鉴于此，本申请提供一种新的技术方案，可以解决相关技术中无法准确识别特征账号的技术问题。
为实现上述目的，本申请提供技术方案如下：
根据本发明的第一方面，提出了一种识别特征账号的方法，包括：
选取历史交易数据中的首活转账交易记录，其中首活转账交易记录对应的被转账方账号由转账方账号的转账操作而被首次激活；
将所述首活转账交易记录对应的转账方账号与被转账方账号之间的关联关系映射为父级节点与子级节点之间的对应关系，并建立对应的树状关系网络；
当所述树状关系网络满足预设特征条件时，判定该树状关系网络中包含特征账号。
根据本发明的第二方面，提出了一种识别特征账号的装置，包括：
记录选取单元，选取历史交易数据中的首活转账交易记录，其中首活转账交易记录对应的被转账方账号由转账方账号的转账操作而被首次激活；
网络建立单元，将所述首活转账交易记录对应的转账方账号与被转账方账号之间的关联关系映射为父级节点与子级节点之间的对应关系，并建立对 应的树状关系网络；
判定单元，当所述树状关系网络满足预设特征条件时，判定该树状关系网络中包含特征账号。
由以上技术方案可见，本申请通过对特定类型的用户账号交互数据的分析，基于特定类型的用户账号交互数据对应的数据流动链路特征，可以实现对特征账号的准确识别。本申请还通过将基于特定类型的用户账号交互数据的账号关系转换为树状关系网络，便于执行账号间的关系识别，并有助于提升对特征账号进行识别时的准确性。
附图说明
图1示出了根据本发明的一示例性实施例的识别特征账号的方法的示意流程图；
图2示出了根据本发明的一示例性实施例的节点关系的示意图；
图3示出了根据本发明的一示例性实施例的通过消息传递生成树状关系网络的示意流程图；
图4示出了根据本发明的一示例性实施例的对节点进行初始化赋值的示意图；
图5示出了根据本发明的一示例性实施例的通过消息传递生成树状关系网络的示意图；
图6示出了根据本发明的一示例性实施例的树状关系网络中的资金传递的示意图；
图7示出了根据本发明的一示例性实施例的服务器的结构示意图；
图8示出了根据本发明的一示例性实施例的识别特征账号的装置的示意框图。
具体实施方式
本申请通过对特定类型的用户账号交互数据的分析，确定在相应的用户 账号间形成的数据流动链，并通过将该数据流动链生成为树状关系网络，从而基于该树状关系网络的特征，识别其中的特征账号。为对本申请进行进一步说明，提供下列实施例：
请参考图1，图1示出了根据本发明的一示例性实施例的识别特征账号的方法，可以应用于服务器，该方法包括：
步骤102，选取历史行为数据中的特定类型的用户账号交互数据，该特定类型的用户账号交互数据对应的数据接收方账号接收到的来自数据发送方账号的交互信息为所述数据接收方账号在注册后接收到的首条交互信息；
在本实施例中，特定类型的用户账号交互数据体现出相应的数据接收方账号与数据发送方账号之间的数据流动特征，基于该数据流动特征的分析，即可确定相应的数据交互双方是否为特征账号。其中，针对不同的数据交互场景，用户账号之间的交互数据的类型可以有很多，比如在即时通讯(IM，Instant Messaging)的应用场景中，交互数据可以为用户账号之间传输的即时通讯消息，比如在社交网络(SNS，Social Network Sites)的应用场景中，交互数据可以为用户账号之间传送的通讯消息，比如在电商交易场景中，交互数据可以为用户账号之间的转账交易信息。
步骤104，将所述特定类型的用户账号交互数据对应的数据发送方账号与数据接收方账号之间的关联关系映射为父级节点与子级节点之间的对应关系，并建立对应的树状关系网络；
在本实施例在，由于每个数据发送方账号可以通过数据交互操作而“首活(由于是数据接收方账号在注册后接收到的首条交互消息，因而可以认为该消息“首活”或“首次激活”了相应的数据接收方账号)”多个数据接收方账号，而每个被首活的数据接收方账号同样可以用于首活其他的账号，因而本申请即可通过该“一对多”(一个数据发送方账号对应多个被首活的数据接收方账号)的对应关系和关系传递(比如B账号被A账号首活，而B账号也可以首活C账号，从而使得“首活”关系在A-B-C之间传递)的特点，由多个特定类型的用户账号交互数据生成对应的树状关系网络，且该树状关系网络 中包含各账号之间进行数据交互操作时的数据流动关系。
步骤106，当所述树状关系网络满足预设特征条件时，判定该树状关系网络中包含特征账号。
由上述实施例可知，本申请通过分析特定类型的用户账号交互数据，利用首活关系在账号之间的传递特点，将每对账号之间的首活关系转换为树状关系网络，从而通过对该树状关系网络中包含的数据流动特征，能够准确分析出该树状关系网络中是否包含特征账号。
作为一示例性实施例，本申请在构建树状关系网络时，可以采用并行计算模型来实现，即通过并行计算来提升该树状关系网络的建立效率。当然，本领域技术人员应该理解的是，本申请并不对树状关系网络的建立方式进行限制，显然也可以通过其他方式实现树状关系网络的建立。
其中，当采用并行计算模型建立树状关系网络时，可以优选使用BSP(Bulk Synchronous Parallel Computing Model，整体同步并行计算模型)建立所述树状关系网络。下面对基于BSP模型建立树状关系网络的过程进行详细描述。
请参考图2，假定提取历史行为数据中的特定类型的用户账号交互数据，并通过对该用户账号交互数据的分析，得到图2所示的账号之间的首活关系，比如图2(a)所示的由节点A首活节点B、由节点A首活节点C，图2(b)所示的由节点C首活节点D，图2(c)所示的由节点D首活节点E、由节点D首活节点F。虽然从图2可以轻易得出多个节点之间的树状关系，但基于数量巨大的特定类型的用户账号交互数据，服务器得到的账号之间的首活关系的数量也非常大，则服务器只能够了解到每对账号之间是否存在首活关系，但并不能够容易得出整个树状关系网络。
请参考图3，图3示出了根据本发明的一示例性实施例的通过消息传递生成树状关系网络的示意流程，包括：
步骤302，初始化操作：将特定类型的用户账号交互数据对应的所有账 号映射为对应的节点，每个节点配置有对应的唯一标识和唯一值。比如对于节点A、节点B、节点C等，相应的唯一标识即A、B、C等；而每个节点的唯一值的形式可以为：自身ID+“#”+superstep(超步)步数，其中的ID为每个节点的唯一标识，比如上述的A、B、C等，superstep是指每个迭代操作的步骤，而“#”则用于区分“ID”和“superstep”的值。可见，当ID和superstep的步数容易区分时，也可以省去“#”。
基于上述赋值规则，如图4所示，节点A对应的唯一值为A#0、节点B对应的唯一值为B#0、节点C对应的唯一值为C#0、节点D对应的唯一值为D#0、节点E对应的唯一值为E#0、节点F对应的唯一值为F#0，然后通过各节点对应值，实现在每个superstep中的信息传递。
步骤304，当superstep＝0时，每个节点直接将包含自身的唯一标识的消息传递至下游节点。其中，下游节点即直接以单向边与本节点相连的节点，比如图5所示：在图5(a1)中，节点A将“A”直接传递至节点B和节点C，节点B不存在下游节点；在图5(b1)中，节点C将“C”直接传递至节点D；在图5(c1)中，节点D直接将“D”传递至节点E和节点F，节点E和节点F不存在下游节点。
在下述的步骤306至步骤320中，需要通过多次迭代，实现对节点上消息的传递，则每次迭代都将执行步骤306至步骤320，下面结合图5进行详细说明。
当superstep＝1时，迭代操作包括：
步骤306，每个节点判断是否接收到消息。对于superstep＝1的情况下，即判断每个节点是否接收到superstep＝0时，由上游节点传递的消息，若接收到，则转入步骤308，否则转入步骤314。
具体地，如图5所示：当superstep＝1时，对应于步骤304中的消息传递，如图5(a1)所示，节点B和节点C接收到来自节点A的消息“A”；如图5(b1)所示，节点D接收到来自节点C的消息“C”；如图5(c1)所示，节点E和节点F接收到来自节点D的消息“D”；同时，节点A未接收到消息。
步骤308，对于在步骤306中接收到消息的节点，将自身的唯一值更新为：接收到的消息中的ID值+“#”+当前superstep。
具体地，如图5所示：由于在图5(a1)中，节点B和节点C接收到来自节点A的“A”，而此时处于superstep＝1的阶段，因而由图5(a1)转入图5(a2)，即节点B和节点C的唯一值均更新为“A#1”；类似地，如图5(b2)所示：节点D的唯一值更新为“C#1”，如图5(c2)所示：节点E和节点F的唯一值更新为“D#1”。
步骤310，判断在步骤308完成自身的唯一值的更新的节点，是否存在对应的下游节点，若存在则转入步骤312，否则转入步骤314。
具体地，如图5所示：在图5(a2)中，节点B不存在下游节点；在图5(b2)中，节点C存在下游节点为节点D；在图5(c2)中，节点D存在下游节点为节点E和节点F，且节点E和节点F不存在下游节点。
步骤312，步骤310中判定存在下游节点的节点，将自身接收到的消息继续传递至下游节点。
具体地，如图5所示：在图5(b2)中，节点C将消息“A”传递至节点D；在图5(c2)中，节点D将消息“C”传递至节点E和节点F。
步骤314，对于步骤306中未接收到消息的节点，或步骤310中不存在下游节点的节点，在当前superstep中结束对该节点的处理。具体地，节点A未接收到消息，而节点B、节点E和节点F不存在下游节点。
步骤316，判断是否所有的节点之间均未有消息发送，或者superstep的迭代次数是否已达到最大迭代次数，若未有消息发送或达到最大迭代次数，则转入步骤318，否则转入步骤320。
步骤318，结束并输出每个节点的唯一值。此时尚未满足转入步骤318的条件。
步骤320，进入下一个superstep，即superstep＝2，并返回步骤306。
当superstep＝2时，迭代操作包括：
步骤306’(在superstep＝2时，通过“步骤306’”来区分“步骤306”，其余 步骤类似)，每个节点判断是否接收到消息。对于superstep＝2的情况下，即判断每个节点是否接收到superstep＝1时，由上游节点传递的消息，若接收到，则转入步骤308，否则转入步骤314。
具体地，如图5所示：当superstep＝1时，对应于步骤312中的消息传递，如图5(b2)所示，节点D接收到来自节点C的消息“A”；如图5(c2)所示，节点E和节点F接收到来自节点D的消息“C”；同时，节点A、节点B和节点C未接收到消息。
步骤308’，对于在步骤306’中接收到消息的节点，将自身的值更新为：接收到的消息中的ID值+“#”+当前superstep。
具体地，如图5所示：在图5(b3)中，节点D的唯一值更新为“A#2”，在图5(c3)中，节点E和节点F的唯一值更新为“C#2”。
步骤310’，判断在步骤308’完成自身的唯一值的更新的节点，是否存在对应的下游节点，若存在则转入步骤312’，否则转入步骤314’。
具体地，如图5所示：在图5(c3)中，节点D存在下游节点为节点E和节点F，节点E和节点F不存在下游节点。
步骤312’，步骤310’中判定存在下游节点的节点，将自身的值传递至下游节点。
具体地，如图5所示：在图5(c3)中，节点D将“A”传递至节点E和节点F。
步骤314’，对于步骤306’中未接收到消息的节点，或步骤310’中不存在下游节点的节点，在当前superstep中结束对该节点的处理。具体地，节点A、节点B、节点C未接收到消息，而节点E和节点F不存在下游节点。
步骤316’，判断是否所有的节点之间均未有消息发送，或者superstep的迭代次数是否已达到最大迭代次数，若未有消息发送或达到最大迭代次数，则转入步骤318’，否则转入步骤320’。
步骤318’，结束并输出每个节点的唯一值。此时尚未满足转入步骤318’的条件。
步骤320’，进入下一个superstep，即superstep＝3，并返回步骤306’。
当superstep＝3时，迭代操作包括：
步骤306”(在superstep＝3时，通过“步骤306””来区分“步骤306”和“步骤306’”，其余步骤类似)，每个节点判断是否接收到消息。对于superstep＝3的情况下，即判断每个节点是否接收到superstep＝2时，由上游节点传递的消息，若接收到，则转入步骤308”，否则转入步骤314”。
具体地，如图5所示：当superstep＝2时，对应于步骤312’中的消息传递，如图5(c3)所示，节点E和节点F接收到来自节点D的消息“A”；同时，节点A、节点B、节点C和节点D未接收到消息。
步骤308”，对于在步骤306”中接收到消息的节点，将自身的唯一值更新为：接收到的消息中的ID+“#”+当前superstep。
具体地，如图5所示：在图5(c4)中，节点E和节点F的唯一值更新为“A#3”。
步骤310”，判断在步骤308”完成自身的值的更新的节点，是否存在对应的下游节点，若存在则转入步骤312”，否则转入步骤314”。
具体地，如图5所示：在图5(c4)中，节点E和节点F不存在下游节点，直接转入步骤314”。
步骤314”，对于步骤306”中未接收到消息的节点，或步骤310”中不存在下游节点的节点，在当前superstep中结束对该节点的处理。具体地，节点A、节点B、节点C和节点D未接收到消息，而节点E和节点F不存在下游节点。
步骤316”，判断是否所有的节点之间均未有消息发送，或者superstep的迭代次数是否已达到最大迭代次数，若未有消息发送或达到最大迭代次数，则转入步骤318”，否则转入步骤320”。由于在此superstep中，所有节点均未有消息发送，因而转入步骤318”。
步骤318”，结束并输出每个节点的唯一值。
具体地，如图5所示：在图5(a4)中，节点A的唯一值为A#0，节点 B的唯一值为A#1，节点C的唯一值为A#1；在图5(b4)中，节点D的唯一值为A#2；在图5(c4)中，节点E和节点F的唯一值均为A#3。
服务器根据获取每个节点输出的唯一值，了解到存在以节点A为根节点的树状关系网络，且节点B和节点C为节点A的一级子节点、节点D为节点A的二级子节点、节点E和节点F为节点A的三级子节点，并形成图5(d)所示的树状关系网络。
通过图3所示的流程或其他技术手段，可以得到树状关系网络，则对于每个树状关系网络，均可以根据该网络的特征，判断其中是否包含特征账号。
作为一示例性实施例，可以根据每个树状关系网络中包含节点的数量进行判断。具体地，由于在基于正常交易构成的树状关系网络中，节点之间的关系比较简单，不会存在很多节点在多个层级上相互关联的情况，因而可以获取每个树状关系网络中包含的节点数量，则当节点数量大于或等于预设数量阈值的情况下，即可判定相应的树状关系网络中包含有特征账号对应的节点。
作为另一示例性实施例，可以根据每个树状关系网络的最大树生长速度进行判断。具体地，由于在基于正常交易构成的树状关系网络中，树生长(即节点增加)的速度较均匀，不会具有强聚集性，因而可以获取每个树状关系网络在单位时间长度内的最大节点增加速度，比如最大单日(即单位时间长度为每日)节点增加数量，则当最大节点增加速度大于或等于预设速度阈值的情况下，即可判定相应的树状关系网络中包含有特征账号对应的节点。
本申请可以应用于各种类型的数据交互场景，下面以其中的典型应用为例，对本申请的技术方案进行详细说明。
1)电商平台
当本申请的技术方案应用于电商平台时，特定类型的用户账号交互数据可以为用户账号之间的转账交易信息。比如对于“支付宝”平台，“特征账号” 即存在虚假交易的账号，比如卖家用户自己或雇佣第三方控制一系列虚拟买家账号，并通过这些虚拟买家账号在支付宝平台上进行虚假交易，从而造成卖家用户的信誉虚高、商品排名上升，不利于买家用户做出正确判断；另外某恶意用户控制一系列虚拟买家账号，通过这些虚拟买家账号恶意骗取卖方商户的营销资源，非法获取不当利润。而基于本申请的技术方案，即可有效识别上述虚假交易涉及的虚拟买家账号，从而对其进行搜索、购买等权限的权重做出合理调整。
在电商平台的应用场景下，除了上述根据树状关系网络中的节点数量或增长速度外，还可以根据树状关系网络中的节点之间的转账数额进行判断是否包含特征账号。其中，在电商平台的应用场景下，特定类型的用户账号交互数据具体可以为首活转账交易记录，即该首活转账交易记录对应的被转账方账号由转账方账号的转账操作而被首次激活。
比如在图6所述的树状关系网络中，获取涉及的所有首活转账交易记录对应的转账数额，并将对应的转账数额之间的差值小于或等于预设差值阈值的首活转账交易记录作为同一分组。假定预设差值阈值为“0.5万”，则由于节点a与节点b之间的转账数额为20万、节点b与节点e之间的转账数额为19.8万、节点e与节点i之间的转账数额为20.2万、节点i与节点k之间的转账数额为20万、节点k与节点n之间的转账数额为19.8万，则确定对应的5条首活转账交易记录为一组；由于节点b与节点d之间的转账数额为3万、节点e与节点h之间的转账数额为3万，则确定对应的2条首活转账交易记录为一组；同时，由于节点e与节点j之间的转账数额为6万、节点j与节点l之间的转账数额为5.8确定对应的2条首活转账交易记录为一组；而其他转账数额对应的首活转账交易记录各自形成一组，具体情况如表1所示：
Figure PCTCN2015086617-appb-000001
表1
根据本申请的一示例性实施例，当存在至少一个分组内包含的首活转账交易记录的数量在所有首活转账交易记录中所占比例大于或等于预设比例阈值时，确定该树状关系网络中包含特征账号。假定该预设比例阈值为30％，则由于分组1所占比例为41.7％＞30％，确定图6所示的树状关系网络中包含特征账号。
进一步地，可以将满足“所占比例大于或等于预设比例阈值”的分组内包含的首活转账交易记录对应的被转账方账号判定为特征账号，比如分组1中的节点b、节点e、节点i、节点k和节点n，可以确定节点a对应的卖家账号通过这些节点对应的账号执行虚假交易，即这些账号为节点a对应的卖家账号采用的虚拟买家账号。因此，可以限制特征账号的预设权限；同时，也可以对节点a对应的卖家账号进行处理，比如通过特征账号完成的交易不被用于计算相应卖家的信誉、出货量等信息。
需要说明的是，根据得到的树状关系网络，可以仅根据其中包含的节点 数量、最大节点增加速度或每组首活转账交易记录所占的比例，也可以通过结合上述三种手段，以确定该网络中是否包含特征账号；在确定树状关系网络中包含特征账号后，即可根据对首活转账交易记录的分组情况，具体确定特征账号。
同时，可以根据实际需求调整“预设数量阈值”、“预设差值阈值”、“预设比例阈值”、“预设速度阈值”的数值，以控制对特征账号进行选取的严格程度。具体地，当预设数量阈值越大、预设差值阈值越大、预设比例阈值越大、预设速度阈值越大时，相应的选取标准越严格(即更难被确定存在特征账号)，反之则越宽松(即更容易被确定存在特征账号)。
2)即时通讯或社交网络
当本申请的技术方案应用于即时通讯或社交网络场景时，特定类型的用户账号交互数据可以为用户账号之间的即时通讯消息或账号关联消息。其中，通讯消息包括即时通讯消息和非即时通讯消息，比如留言、评论等；账号关联消息包括好友添加请求、群加入请求、群成员邀请请求、账号关注操作等。具体地，“特征账号”即存在欺诈或骚扰行为的账号，比如用户通过控制一系列“小号”，并通过这些“小号”对其他用户进行欺诈或骚扰，比如发送欺骗消息、木马网站的网址或广告等。而基于本申请的技术方案，即可有效识别上述欺诈或骚扰行为涉及的账号，从而对其进行消息发送、好友添加等权限的权重做出合理调整。
对应于上述的识别特征账号的方法，本申请还提出了图7所示的根据本申请的一示例性实施例的服务器的示意结构图。请参考图15，在硬件层面，该服务器包括处理器、内部总线、网络接口、内存以及非易失性存储器，当然还可能包括其他业务所需要的硬件。处理器从非易失性存储器中读取对应的计算机程序到内存中然后运行，在逻辑层面上形成识别特征账号的装置。当然，除了软件实现方式之外，本申请并不排除其他实现方式，比如逻辑器件抑或软硬件结合的方式等等，也就是说以下处理流程的执行主体并不限定 于各个逻辑单元，也可以是硬件或逻辑器件。
请参考图8，在软件实施方式中，该识别特征账号的装置可以包括记录选取单元、网络建立单元和判定单元。其中：
记录选取单元，选取历史行为数据中的特定类型的用户账号交互数据，该特定类型的用户账号交互数据对应的数据接收方账号接收到的来自数据发送方账号的交互信息为所述数据接收方账号在注册后接收到的首条交互信息；
网络建立单元，将所述特定类型的用户账号交互数据对应的数据发送方账号与数据接收方账号之间的关联关系映射为父级节点与子级节点之间的对应关系，并建立对应的树状关系网络；
判定单元，当所述树状关系网络满足预设特征条件时，判定该树状关系网络中包含特征账号。
可选的，所述网络建立单元具体用于：
分别为每个节点生成对应的传递消息，该传递消息中包含相应节点的唯一标识；
根据所述对应关系执行迭代操作，其中在每次迭代操作中：当不存在父级节点时，相应的每个节点将本地的传递消息发送至对应的子级节点；当存在父级节点时，相应的每个节点接收来自对应的父级节点的传递消息，将接收到的传递消息中包含的唯一标识和当前迭代次数的线性组合值作为相应节点的唯一值，并将接收到的传递消息发送至对应的子级节点；当不存在对应的子级节点或已达到预设的迭代次数时，每个节点上报其唯一值；
其中，由上报的唯一值中包含的唯一标识相同的节点构成对应的树状关系网络，且每个节点在该树状关系网络中的层级状态对应于上报的唯一值中包含的当前迭代次数。
可选的：所述网络建立单元通过并行计算模型建立所述树状关系网络。
可选的：所述网络建立单元利用整体同步并行计算BSP模型建立所述树状关系网络。
可选的，所述预设特征条件包括：
所述树状关系网络中包含节点的数量大于或等于预设数量阈值。
可选的，所述预设特征条件包括：
所述树状关系网络的最大节点增加速度大于或等于预设速度阈值。
可选的，所述预设特征条件包括：
在所述树状关系网络对应的所有特定类型的用户账号交互数据中，将对应的转账数额之间的差值小于或等于预设差值阈值的特定类型的用户账号交互数据作为同一分组时，存在至少一个分组内包含的特定类型的用户账号交互数据的数量在所有特定类型的用户账号交互数据中所占比例大于或等于预设比例阈值。
可选的，所述判定单元具体用于：
将所述至少一个分组内包含的特定类型的用户账号交互数据对应的数据接收方账号判定为所述特征账号，并限制所述特征账号的预设权限。
因此，本申请通过对特定类型的用户账号交互数据的分析，确定在相应的用户账号间形成的数据流动链，并通过将该数据流动链生成为树状关系网络，从而基于该树状关系网络的特征，识别其中的特征账号。
在一个典型的配置中，计算设备包括一个或多个处理器(CPU)、输入/输出接口、网络接口和内存。
内存可能包括计算机可读介质中的非永久性存储器，随机存取存储器(RAM)和/或非易失性内存等形式，如只读存储器(ROM)或闪存(flash RAM)。内存是计算机可读介质的示例。
计算机可读介质包括永久性和非永久性、可移动和非可移动媒体可以由任何方法或技术来实现信息存储。信息可以是计算机可读指令、数据结构、程序的模块或其他数据。计算机的存储介质的例子包括，但不限于相变内存(PRAM)、静态随机存取存储器(SRAM)、动态随机存取存储器(DRAM)、其他类型的随机存取存储器(RAM)、只读存储器(ROM)、电可擦除可编程只读存储器(EEPROM)、快闪记忆体或其他内存技术、只读光盘只读存储器(CD-ROM)、数字多功能光盘(DVD)或其他光学存储、磁盒式磁带，磁带磁磁盘存储或其他磁 性存储设备或任何其他非传输介质，可用于存储可以被计算设备访问的信息。按照本文中的界定，计算机可读介质不包括暂存电脑可读媒体(transitory media)，如调制的数据信号和载波。
还需要说明的是，术语“包括”、“包含”或者其任何其他变体意在涵盖非排他性的包含，从而使得包括一系列要素的过程、方法、商品或者设备不仅包括那些要素，而且还包括没有明确列出的其他要素，或者是还包括为这种过程、方法、商品或者设备所固有的要素。在没有更多限制的情况下，由语句“包括一个……”限定的要素，并不排除在包括所述要素的过程、方法、商品或者设备中还存在另外的相同要素。
以上所述仅为本申请的较佳实施例而已，并不用以限制本申请，凡在本申请的精神和原则之内，所做的任何修改、等同替换、改进等，均应包含在本申请保护的范围之内。

Claims (16)

  1. 一种识别特征账号的方法，其特征在于，包括：
    选取历史行为数据中的特定类型的用户账号交互数据，该特定类型的用户账号交互数据对应的数据接收方账号接收到的来自数据发送方账号的交互信息为所述数据接收方账号在注册后接收到的首条交互信息；
    将所述特定类型的用户账号交互数据对应的数据发送方账号与数据接收方账号之间的关联关系映射为父级节点与子级节点之间的对应关系，并建立对应的树状关系网络；
    当所述树状关系网络满足预设特征条件时，判定该树状关系网络中包含特征账号。
  2. 根据权利要求1所述的方法，其特征在于，建立对应的树状关系网络，具体包括：
    分别为每个节点生成对应的传递消息，该传递消息中包含相应节点的唯一标识；
    根据所述对应关系执行迭代操作，其中在每次迭代操作中：当不存在父级节点时，相应的每个节点将本地的传递消息发送至对应的子级节点；当存在父级节点时，相应的每个节点接收来自对应的父级节点的传递消息，将接收到的传递消息中包含的唯一标识和当前迭代次数的线性组合值作为相应节点的唯一值，并将接收到的传递消息发送至对应的子级节点；当不存在对应的子级节点或已达到预设的迭代次数时，每个节点上报其唯一值；
    其中，由上报的唯一值中包含的唯一标识相同的节点构成对应的树状关系网络，且每个节点在该树状关系网络中的层级状态对应于上报的唯一值中包含的当前迭代次数。
  3. 根据权利要求1所述的方法，其特征在于：通过并行计算模型建立所述树状关系网络。
  4. 根据权利要求3所述的方法，其特征在于，通过并行计算模型建立所 述树状关系网络，具体包括：
    利用整体同步并行计算BSP模型建立所述树状关系网络。
  5. 根据权利要求1所述的方法，其特征在于，所述预设特征条件包括：
    所述树状关系网络中包含节点的数量大于或等于预设数量阈值。
  6. 根据权利要求1所述的方法，其特征在于，所述预设特征条件包括：
    所述树状关系网络的最大节点增加速度大于或等于预设速度阈值。
  7. 根据权利要求1所述的方法，其特征在于，所述预设特征条件包括：
    在所述树状关系网络对应的所有特定类型的用户账号交互数据中，将对应的转账数额之间的差值小于或等于预设差值阈值的特定类型的用户账号交互数据作为同一分组时，存在至少一个分组内包含的特定类型的用户账号交互数据的数量在所有特定类型的用户账号交互数据中所占比例大于或等于预设比例阈值。
  8. 根据权利要求7所述的方法，其特征在于，还包括：
    将所述至少一个分组内包含的特定类型的用户账号交互数据对应的数据接收方账号判定为所述特征账号，并限制所述特征账号的预设权限。
  9. 一种识别特征账号的装置，其特征在于，包括：
    记录选取单元，选取历史行为数据中的特定类型的用户账号交互数据，该特定类型的用户账号交互数据对应的数据接收方账号接收到的来自数据发送方账号的交互信息为所述数据接收方账号在注册后接收到的首条交互信息；
    网络建立单元，将所述特定类型的用户账号交互数据对应的数据发送方账号与数据接收方账号之间的关联关系映射为父级节点与子级节点之间的对应关系，并建立对应的树状关系网络；
    判定单元，当所述树状关系网络满足预设特征条件时，判定该树状关系网络中包含特征账号。
  10. 根据权利要求9所述的装置，其特征在于，所述网络建立单元具体用于：
    分别为每个节点生成对应的传递消息，该传递消息中包含相应节点的唯 一标识；
    根据所述对应关系执行迭代操作，其中在每次迭代操作中：当不存在父级节点时，相应的每个节点将本地的传递消息发送至对应的子级节点；当存在父级节点时，相应的每个节点接收来自对应的父级节点的传递消息，将接收到的传递消息中包含的唯一标识和当前迭代次数的线性组合值作为相应节点的唯一值，并将接收到的传递消息发送至对应的子级节点；当不存在对应的子级节点或已达到预设的迭代次数时，每个节点上报其唯一值；
    其中，由上报的唯一值中包含的唯一标识相同的节点构成对应的树状关系网络，且每个节点在该树状关系网络中的层级状态对应于上报的唯一值中包含的当前迭代次数。
  11. 根据权利要求9所述的装置，其特征在于：所述网络建立单元通过并行计算模型建立所述树状关系网络。
  12. 根据权利要求11所述的装置，其特征在于：所述网络建立单元利用整体同步并行计算BSP模型建立所述树状关系网络。
  13. 根据权利要求9所述的装置，其特征在于，所述预设特征条件包括：
    所述树状关系网络中包含节点的数量大于或等于预设数量阈值。
  14. 根据权利要求9所述的装置，其特征在于，所述预设特征条件包括：
    所述树状关系网络的最大节点增加速度大于或等于预设速度阈值。
  15. 根据权利要求9所述的装置，其特征在于，所述预设特征条件包括：
    在所述树状关系网络对应的所有特定类型的用户账号交互数据中，将对应的转账数额之间的差值小于或等于预设差值阈值的特定类型的用户账号交互数据作为同一分组时，存在至少一个分组内包含的特定类型的用户账号交互数据的数量在所有特定类型的用户账号交互数据中所占比例大于或等于预设比例阈值。
  16. 根据权利要求15所述的装置，其特征在于，所述判定单元具体用于：
    将所述至少一个分组内包含的特定类型的用户账号交互数据对应的数据接收方账号判定为所述特征账号，并限制所述特征账号的预设权限。
PCT/CN2015/086617 2014-08-27 2015-08-11 识别特征账号的方法及装置 WO2016029794A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201410429738.X 2014-08-27
CN201410429738.XA CN105447028B (zh) 2014-08-27 2014-08-27 识别特征账号的方法及装置

Publications (1)

Publication Number Publication Date
WO2016029794A1 true WO2016029794A1 (zh) 2016-03-03

Family

ID=55398739

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2015/086617 WO2016029794A1 (zh) 2014-08-27 2015-08-11 识别特征账号的方法及装置

Country Status (2)

Country Link
CN (1) CN105447028B (zh)
WO (1) WO2016029794A1 (zh)

Cited By (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2019019385A1 (zh) * 2017-07-26 2019-01-31 平安科技（深圳）有限公司 跨平台数据匹配方法、装置、计算机设备和存储介质
CN109522342A (zh) * 2018-11-30 2019-03-26 北京百度网讯科技有限公司 公安警务管理方法、装置、设备和存储介质
CN109685094A (zh) * 2018-09-21 2019-04-26 阿里巴巴集团控股有限公司 一种基于网络指纹的身份识别方法和装置
CN110032665A (zh) * 2019-03-25 2019-07-19 阿里巴巴集团控股有限公司 确定关系网络图中图节点向量的方法及装置
CN110210883A (zh) * 2018-05-09 2019-09-06 腾讯科技（深圳）有限公司 群控账号识别方法、装置、服务器及存储介质
CN110825919A (zh) * 2018-07-23 2020-02-21 阿里巴巴集团控股有限公司 Id数据处理方法和装置
CN110929049A (zh) * 2019-12-02 2020-03-27 北京明略软件系统有限公司 一种用户账号识别方法及装置
CN111046894A (zh) * 2018-10-15 2020-04-21 北京京东尚科信息技术有限公司 识别马甲账号的方法和装置
CN111641523A (zh) * 2020-05-25 2020-09-08 中国建设银行股份有限公司 用户数据管理方法、装置、系统及存储介质
CN112651433A (zh) * 2020-12-17 2021-04-13 广州锦行网络科技有限公司 一种特权账号异常行为分析方法

Families Citing this family (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107733767B (zh) * 2016-08-08 2020-03-17 腾讯科技（深圳）有限公司 一种社交关系网的建立方法、装置和系统
CN109936525B (zh) 2017-12-15 2020-07-31 阿里巴巴集团控股有限公司 一种基于图结构模型的异常账号防控方法、装置以及设备
CN110619070B (zh) * 2018-06-04 2022-05-10 北京百度网讯科技有限公司 文章生成方法和装置
CN112492534B (zh) * 2020-11-25 2022-04-15 中国联合网络通信集团有限公司 消息处理方法、装置及设备

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20080103798A1 (en) * 2006-10-25 2008-05-01 Domenikos Steven D Identity Protection
CN103309885A (zh) * 2012-03-13 2013-09-18 阿里巴巴集团控股有限公司 一种在电子交易平台中识别特征用户的方法及装置和搜索方法及装置
CN103577991A (zh) * 2012-08-03 2014-02-12 阿里巴巴集团控股有限公司 一种用户的识别方法和装置
CN103778151A (zh) * 2012-10-23 2014-05-07 阿里巴巴集团控股有限公司 一种识别特征群体的方法及装置和搜索方法及装置
CN103955833A (zh) * 2014-03-31 2014-07-30 浙江工商大学 基于虚假交易和社交关系矩阵分析的水军身份确认方法

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102571485B (zh) * 2011-12-14 2014-08-27 上海交通大学 一种在微博平台上识别机器人用户的方法
CN103927307B (zh) * 2013-01-11 2017-03-01 阿里巴巴集团控股有限公司 一种识别网站用户的方法和装置

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20080103798A1 (en) * 2006-10-25 2008-05-01 Domenikos Steven D Identity Protection
CN103309885A (zh) * 2012-03-13 2013-09-18 阿里巴巴集团控股有限公司 一种在电子交易平台中识别特征用户的方法及装置和搜索方法及装置
CN103577991A (zh) * 2012-08-03 2014-02-12 阿里巴巴集团控股有限公司 一种用户的识别方法和装置
CN103778151A (zh) * 2012-10-23 2014-05-07 阿里巴巴集团控股有限公司 一种识别特征群体的方法及装置和搜索方法及装置
CN103955833A (zh) * 2014-03-31 2014-07-30 浙江工商大学 基于虚假交易和社交关系矩阵分析的水军身份确认方法

Cited By (17)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US11030265B2 (en) 2017-07-26 2021-06-08 Ping An Technology (Shenzhen) Co., Ltd. Cross-platform data matching method and apparatus, computer device and storage medium
WO2019019385A1 (zh) * 2017-07-26 2019-01-31 平安科技（深圳）有限公司 跨平台数据匹配方法、装置、计算机设备和存储介质
CN110210883B (zh) * 2018-05-09 2023-08-22 腾讯科技（深圳）有限公司 群控账号识别方法、装置、服务器及存储介质
CN110210883A (zh) * 2018-05-09 2019-09-06 腾讯科技（深圳）有限公司 群控账号识别方法、装置、服务器及存储介质
CN110825919B (zh) * 2018-07-23 2023-05-26 阿里巴巴集团控股有限公司 Id数据处理方法和装置
CN110825919A (zh) * 2018-07-23 2020-02-21 阿里巴巴集团控股有限公司 Id数据处理方法和装置
CN109685094A (zh) * 2018-09-21 2019-04-26 阿里巴巴集团控股有限公司 一种基于网络指纹的身份识别方法和装置
CN111046894A (zh) * 2018-10-15 2020-04-21 北京京东尚科信息技术有限公司 识别马甲账号的方法和装置
CN109522342A (zh) * 2018-11-30 2019-03-26 北京百度网讯科技有限公司 公安警务管理方法、装置、设备和存储介质
CN110032665B (zh) * 2019-03-25 2023-11-17 创新先进技术有限公司 确定关系网络图中图节点向量的方法及装置
CN110032665A (zh) * 2019-03-25 2019-07-19 阿里巴巴集团控股有限公司 确定关系网络图中图节点向量的方法及装置
CN110929049A (zh) * 2019-12-02 2020-03-27 北京明略软件系统有限公司 一种用户账号识别方法及装置
CN110929049B (zh) * 2019-12-02 2023-05-26 北京明略软件系统有限公司 一种用户账号识别方法及装置
CN111641523A (zh) * 2020-05-25 2020-09-08 中国建设银行股份有限公司 用户数据管理方法、装置、系统及存储介质
CN111641523B (zh) * 2020-05-25 2023-01-17 中国建设银行股份有限公司 用户数据管理方法、装置、系统及存储介质
CN112651433B (zh) * 2020-12-17 2021-12-14 广州锦行网络科技有限公司 一种特权账号异常行为分析方法
CN112651433A (zh) * 2020-12-17 2021-04-13 广州锦行网络科技有限公司 一种特权账号异常行为分析方法

Also Published As

Publication number Publication date
CN105447028B (zh) 2019-06-28
CN105447028A (zh) 2016-03-30

Similar Documents

Publication Publication Date Title
WO2016029794A1 (zh) 识别特征账号的方法及装置
US11869012B2 (en) Systems, devices, and methods for DLT-based data management platforms and data products
US10255626B2 (en) Methods, devices, and systems for sending and receiving virtual goods
TWI727467B (zh) 聯盟鏈的可信度驗證方法、系統、裝置及設備
TWI709868B (zh) 基於區塊鏈的資料存證方法和裝置
TWI673666B (zh) 資料風險控制的方法及裝置
JP6184954B2 (ja) 自然言語処理に基づく、種々のオブジェクトに対する係数付与
KR20210089682A (ko) 블록 체인을 사용한 영지식 증명 결제
WO2017067394A1 (zh) 客户服务方式的分配方法及系统
TW201832100A (zh) 區塊鏈系統、資料儲存方法及裝置
JP2023164828A (ja) ブロックチェーンを使用する取引口座データ保守
WO2017050188A1 (zh) 一种交易数据处理方法及装置
JP7157798B6 (ja) ブロックチェーンネットワークを介してデータを通信し、格納し、及び処理するためのブロックチェーンベースのシステム及び方法
CN113678155A (zh) 支付转账处理系统
WO2018126344A1 (zh) 一种数据处理方法及相关设备
CN105391594B (zh) 识别特征账号的方法及装置
TWI706362B (zh) 基於區塊鏈的資料處理方法、裝置和伺服器
WO2020125233A1 (zh) 字段更新方法及装置、电子设备
WO2020134627A1 (zh) 联盟链的可信度验证方法、系统、装置及设备
CN106529953B (zh) 一种对业务属性进行风险识别的方法及装置
WO2019076113A1 (zh) 一种信息处理方法、装置及设备
Atkins et al. A cloud service for end-user participation concerning the internet of things
KR20170116022A (ko) 서비스 구현
CN111161006A (zh) 一种区块链信用服务方法、系统及存储介质
WO2016197873A1 (zh) 事务处理方法和系统

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 15836014

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 15836014

Country of ref document: EP

Kind code of ref document: A1