Data processing method, data processing apparatus, medium, device and product

Abstract

Embodiments of the present specification provide a data processing method, a data processing device, a computer readable storage medium, an electronic device and a computer program product. The data processing includes: determining a propagation path of a private message stored by a target node in a graph relationship network, wherein the private message is sent based on a message passing mechanism of a synchronous parallel BSP computing model. Specifically, comparing attributes corresponding to the propagation path and a preset constraint condition to determine partial self-loop edges in all self-loop edges of a reached node in the propagation path that satisfy the constraint condition. Further, according to target behaviors corresponding to the partial self-loop edges, a risk value corresponding to the target node can be determined.

Description

Technical Field

[0001] This specification relates to the field of data processing technology, and in particular to a data processing method, a data processing apparatus, a computer-readable storage medium, an electronic device, and a computer program product. Background Technology

[0002] With the rapid development of information technologies such as cloud computing, big data, and artificial intelligence, graph relational networks (graph structured data) have emerged as a powerful tool for characterizing relationships. When dealing with large datasets (e.g., reaching hundreds of millions of records), the structure of graph relational networks becomes extremely complex.

[0003] Therefore, there is an urgent need in the existing technology for a solution that is based on graph relationship networks for data processing and has high processing efficiency.

[0004] It should be noted that the information disclosed in the background section above is only used to enhance the understanding of the background of this specification, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] This specification provides a data processing method, a data processing apparatus, a computer-readable storage medium, an electronic device, and a computer program product, which can process data based on graph relationship networks with high efficiency.

[0006] Other features and advantages of the embodiments described herein will become apparent from the following detailed description, or may be learned in part by practice of this specification.

[0007] According to one aspect of the embodiments of this specification, a data processing method is provided, the method comprising: for a target node in a graph relation network, determining the propagation path of a private message stored in the target node, wherein the private message is sent by a message passing mechanism based on a synchronous parallel BSP computing model, and the propagation path includes reaching nodes; determining a subset of self-loop edges among all self-loop edges of the reaching nodes by comparing the attributes corresponding to the propagation path and preset constraints; and determining a risk value corresponding to the target node based on the target behaviors corresponding to the subset of self-loop edges.

[0008] In an exemplary embodiment, based on the above scheme, the method further includes: a message passing mechanism based on the BSP computing model to save the private messages corresponding to each node; determining the relationship edges between nodes based on the relationship data between nodes, and determining their self-loop edges based on the usage data of nodes, to obtain the above graph relationship network; wherein the above nodes include: users and media, and the above media includes one or more of devices, certificates and identification numbers.

[0009] In an exemplary embodiment, based on the above scheme, the method further includes: determining a subgraph in the graph relationship network, wherein the subgraph includes: an initiating node corresponding to a risk initiator and an access node corresponding to a risk triggerer, and the access node includes one or more.

[0010] For a target node in a graph relation network, determining the propagation path of the private message stored in the target node includes: determining the starting node in the subgraph as the target node; and determining the path of the private message propagating from the target node to the reaching node as the propagation path.

[0011] In an exemplary embodiment, based on the above scheme, the method further includes: for the relation edges included in the propagation path, determining the actual association time and / or the actual association count between the two nodes connected by the relation edges as the attribute corresponding to the relation edges; and determining the attribute corresponding to the propagation path based on the attribute corresponding to the relation edges included in the propagation path.

[0012] In an exemplary embodiment, based on the above scheme, determining the actual association time and / or actual association count between the two nodes connected by the relationship edge in the propagation path as the attribute corresponding to the relationship edge includes: when the propagation path sequentially includes the target node and the reach node, determining the actual association time and / or actual association count between the target node and the reach node as the attribute corresponding to the relationship edge.

[0013] The above determination of the attributes corresponding to the propagation path based on the attributes corresponding to the relation edges contained in the propagation path includes: determining the attributes corresponding to the relation edges as the attributes corresponding to the propagation path.

[0014] In an exemplary embodiment, based on the above scheme, determining the actual association time and / or actual association count between the two nodes connected by the relationship edge included in the propagation path as the attribute corresponding to the relationship edge includes: when the propagation path sequentially includes the target node, an intermediate node, and the reach node, determining the actual association time and / or actual association count between the target node and the intermediate node as the attribute corresponding to the first relationship edge; and determining the actual association time and / or actual association count between the intermediate node and the reach node as the attribute corresponding to the second relationship edge.

[0015] The above determination of the attributes corresponding to the propagation path based on the attributes corresponding to the relation edges contained in the propagation path includes: determining the attributes corresponding to the propagation path based on the attributes corresponding to the first relation edge and the attributes corresponding to the second relation edge.

[0016] In an exemplary embodiment, based on the above scheme, determining the attribute corresponding to the propagation path according to the attribute corresponding to the first relation edge and the attribute corresponding to the second relation edge includes: determining the actual association time in the attribute of the propagation path according to the actual association time in the attribute corresponding to the first relation edge and the actual association time in the attribute corresponding to the second relation edge; and / or, determining the actual association count in the attribute of the propagation path according to the actual association count in the attribute of the first relation edge and the actual association count in the attribute of the second relation edge.

[0017] In an exemplary embodiment, based on the above scheme, determining the propagation path of the private message stored by the target node in the graph relation network includes: for any relation edge in the subgraph, comparing the attribute corresponding to the relation edge with a preset filtering condition; and, if the attribute corresponding to the relation edge does not meet the filtering condition, filtering out the propagation path where the relation edge is located, wherein the remaining propagation path after filtering is the propagation path of the private message stored by the target node, and comparing the attribute corresponding to it with the above constraint condition.

[0018] In an exemplary embodiment, based on the above scheme, the attributes corresponding to the relation edge include the actual association time and the actual association count, and the filtering conditions include the filtering association time and the filtering association count. The comparison of the attributes corresponding to the relation edge with the preset filtering conditions includes: for a target relation edge, comparing the actual association time and the filtering association time in the attributes corresponding to the target relation edge, and comparing the actual association count and the filtering association count in the attributes corresponding to the target relation edge. Wherein, if the actual association time does not meet the filtering association time, and / or the actual association count does not meet the filtering association count, the propagation path where the target relation edge is located will be filtered out.

[0019] In an exemplary embodiment, based on the above scheme, the attributes corresponding to the propagation path include actual association time and actual association count, and the constraints include constraint association time and constraint association count. The method of determining a partial self-loop edge among all self-loop edges of the reached node by comparing the attributes corresponding to the propagation path with the preset constraints includes: comparing the actual association time in the attributes corresponding to the propagation path with the constraint association time to determine a first target behavior group that satisfies the constraint association time; comparing the actual association count in the attributes corresponding to the propagation path with the constraint association count to determine a second target behavior group that satisfies the constraint association count; and determining the intersection of the first target behavior group and the second target behavior group as the partial self-loop edge, wherein the behavior represented by the partial self-loop edge is a target behavior associated with the private message.

[0020] In an exemplary embodiment, based on the above scheme, determining the risk value corresponding to the target node according to the target behaviors corresponding to the aforementioned partial self-loop edges includes: summing the target behaviors corresponding to the aforementioned partial self-loop edges to obtain the risk value corresponding to the target node; or, determining the weights of the target behaviors corresponding to the aforementioned partial self-loop edges, multiplying the target behaviors by their corresponding weights, and then summing the products to obtain the risk value corresponding to the target node.

[0021] In an exemplary embodiment, based on the above scheme, after determining the risk value corresponding to the target node, the method further includes: generating warning information about the target node when it is determined that the risk value corresponding to the target node is greater than a preset threshold.

[0022] According to another aspect of the embodiments of this specification, a data processing apparatus includes: a path determination module, a self-loop edge filtering module, and a risk value determination module.

[0023] The path determination module is used to determine the propagation path of the private message stored by the target node in the graph relation network, wherein the private message is sent by the message passing mechanism based on the synchronous parallel BSP computing model, and the propagation path includes the reaching node; the self-loop edge filtering module is used to determine some self-loop edges among all self-loop edges of the reaching node by comparing the attributes corresponding to the propagation path and the preset constraints; and the risk value determination module is used to determine the risk value corresponding to the target node according to the target behavior corresponding to the aforementioned self-loop edges.

[0024] According to another aspect of the embodiments of this specification, an electronic device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the data processing method as described in the above embodiments.

[0025] According to another aspect of the embodiments of this specification, a computer-readable storage medium is provided, wherein instructions are stored therein, which, when executed on a computer or processor, cause the computer or processor to perform the data processing method as described in the above embodiments.

[0026] According to one aspect of an embodiment of this specification, a computer program product including instructions is provided that, when the computer program product is run on a computer or processor, causes the computer or processor to perform the data processing method as described in the above embodiment.

[0027] The data processing methods, data processing devices, computer-readable storage media, electronic devices, and computer program products provided in the embodiments of this specification have the following technical effects:

[0028] For a target node in a graph relational network, the propagation path of the private message stored in that target node is determined. This private message is sent using a message passing mechanism based on a synchronous parallel BSP computing model, thereby enabling parallel processing of each node in the graph relational network and improving data processing efficiency. Specifically, the attributes corresponding to the propagation path and preset constraints are compared to identify some self-loop edges that satisfy the constraints among all self-loop edges reaching the node in the propagation path. Further, based on the target behaviors corresponding to these self-loop edges, the risk value corresponding to the target node can be determined. Therefore, in a credit risk control scenario, the embodiments of this specification can efficiently and accurately calculate the contribution of personal fraud risk, thereby effectively reducing financial risk.

[0029] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this specification. Attached Figure Description

[0030] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this specification and, together with the description, serve to explain the principles of this specification. It is obvious that the drawings described below are merely some embodiments of this specification, and those skilled in the art can obtain other drawings based on these drawings without any inventive effort.

[0031] Figure 1 This is a flowchart illustrating the data processing method provided in the embodiments of this specification.

[0032] Figure 2 This is a flowchart illustrating a data processing method provided in another embodiment of this specification.

[0033] Figure 3 This is a flowchart illustrating the method for constructing a graph relationship network provided in the embodiments of this specification.

[0034] Figure 4 This is a schematic diagram of the graph relationship network provided in the embodiments of this specification.

[0035] Figure 5a According to Figure 4 The diagram shows a subgraph determined by the relationship network shown.

[0036] Figure 5b According to Figure 4 The diagram shows another subgraph determined by the relationship network shown.

[0037] Figure 6 This is a schematic flowchart illustrating the relation edge filtering method provided in the embodiments of this specification.

[0038] Figure 7 This is a schematic diagram illustrating the determination of untrusted relation edges in a subgraph, as provided in the embodiments of this specification.

[0039] Figure 8 This is a flowchart illustrating a data processing method provided in another embodiment of this specification.

[0040] Figure 9 This is a flowchart illustrating the method for determining the attributes corresponding to the propagation path provided in the embodiments of this specification.

[0041] Figure 10 This is a schematic diagram of the nodes and their self-loop edges provided in the embodiments of this specification.

[0042] Figure 11 This is a schematic diagram of the data processing apparatus provided in the embodiments of this specification.

[0043] Figure 12 This is a schematic diagram of the structure of the electronic device provided in the embodiments of this specification. Detailed Implementation

[0044] To make the objectives, technical solutions, and advantages of this specification clearer, the embodiments of this specification will be described in further detail below with reference to the accompanying drawings.

[0045] In the following description, when referring to the accompanying drawings, the same numbers in different drawings denote the same or similar elements unless otherwise indicated. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this specification. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this specification as detailed in the appended claims.

[0046] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, these embodiments are provided to make this specification more comprehensive and complete, and to fully convey the concept of example embodiments to those skilled in the art. The described features, structures, or characteristics can be combined in any suitable manner in one or more embodiments. In the following description, numerous specific details are provided to give a full understanding of the embodiments described herein. However, those skilled in the art will recognize that the technical solutions described herein may be practiced with one or more of the specific details omitted, or other methods, components, apparatus, steps, etc., may be employed. In other instances, well-known technical solutions are not shown or described in detail to avoid obscuring various aspects of this specification.

[0047] Furthermore, the accompanying drawings are merely illustrative diagrams of this specification and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and therefore repeated descriptions of them will be omitted. Some block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities may be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor devices and / or microcontroller devices.

[0048] In credit risk control scenarios, if the contribution of individual fraud risk cannot be calculated efficiently and accurately, fraud risk is likely to emerge, which can easily lead to asset losses for the company and also create financial risks. Fraud risk, specifically the risk of malicious deception, is a type of operational risk. For cash loans, this can manifest as multiple identity impersonations, collusion between internal and external parties, and cash-out schemes. For e-commerce, it can manifest as "coupon hunting," fake orders, fake positive reviews, and shipping fee fraud.

[0049] To calculate an individual's contribution to fraud risk (denoted as risk value), related technologies using Structured Query Language (SQL) for calculation or filtering in multi-degree relational networks consume significant amounts of machine resources and memory, with total computation time reaching days. This fails to meet the needs of risk discovery and cannot efficiently determine an individual's contribution to fraud risk.

[0050] The embodiments provided in this specification utilize a massively parallel graph computation method to construct a paradigm for calculating an individual's contribution to fraud risk, reconstructing the spread of fraud risk within the graph for each individual. This allows for the calculation of an individual's risk contribution within minutes (e.g., 15 minutes) on a dataset of hundreds of millions of relationships, assessing their current risk level within the overall population. The solutions provided in these embodiments enable the timely detection of risk anomalies, thereby safeguarding the risk control of online lending.

[0051] Specifically, Figure 1 This is a flowchart illustrating the data processing method provided in the embodiments of this specification. (See reference...) Figure 1 The method shown in the figure includes: S110-S130.

[0052] In S110, for a target node in a graph relation network, the propagation path of the private message stored in the target node is determined, wherein the private message is sent by a message passing mechanism based on a synchronous parallel BSP computing model, and the propagation path includes reaching nodes.

[0053] A relationship network refers to a collection of participants and their relationships, where relational behaviors serve as the edges between people. Common relationship networks include device relationships and kinship relationships. The graph relationship network described above is a relationship network represented in graph form.

[0054] Pregel graph computation mode refers to a message-passing batch processing-based parallel engine centered on graph vertices. The Bulk Synchronous Parallel Computing Model (BSP), also known as the "large synchronization" model, involves a series of global hypersteps (a hyperstep is an iteration in computation). Each hyperstep mainly includes three components: local computation: each participating processor has its own computational task; communication: the processor group exchanges data with each other; and barrier synchronization: when a processor encounters a "roadblock" (or barrier), it waits until all other processors have completed their computation steps.

[0055] In S120, by comparing the attributes corresponding to the propagation path with the preset constraints, some self-loop edges are determined from all self-loop edges of the reached node.

[0056] In S130, the risk value corresponding to the target node is determined based on the target behavior corresponding to the partial self-loop edges.

[0057] This instruction manual Figure 1In the embodiment shown, for a target node in a graph relation network, the propagation path of the private message stored by that target node is determined. This private message is sent using a message passing mechanism based on a synchronous parallel BSP computing model, thereby enabling parallel processing of each node in the graph relation network and improving data processing efficiency. Specifically, the attributes corresponding to the propagation path and preset constraints are compared to identify some self-loop edges that satisfy the constraints among all self-loop edges reaching the node in the propagation path. Further, based on the target behaviors corresponding to these self-loop edges, the risk value corresponding to the target node can be determined. Therefore, in a credit risk control scenario, this embodiment can efficiently and accurately calculate the contribution of personal fraud risk, thereby effectively reducing financial risk.

[0058] In an exemplary embodiment, Figure 2 This is a flowchart illustrating another embodiment of the data processing method provided in this specification. The flowchart describes the overall process of determining a user's risk contribution value in a credit scenario. Specifically, this embodiment uses the Pregel pattern as the computational framework for distributed graph computing. Each participating processor processes the computational task corresponding to the current point, completing one superstep iteration through a single Bulk Synchronous Parallel Computing Model (BSP) computation process. The entire scheme is divided into four stages:

[0059] Step A: Graph Construction Phase. In this phase, a distributed graph relationship network is constructed based on large-scale relational and behavioral data. It should be noted that the relational and behavioral data involved in this specification were obtained with full authorization. Step B: Graph Traversal Phase. In this phase, using the message propagation mechanism based on the BSP calculation model, the graph traverses the subgraph structure of the current risk initiator (corresponding to the aforementioned target node) starting from the risk initiator. Step C: Reach Node Calculation Phase. In this phase, the risk status of each reach node is calculated using the trigger edge as the dimension. Step D: Target Node Risk Value Calculation Phase. In this phase, messages are sent back to the target node through the tail end (i.e., reach node), thereby summarizing the risk contribution of all the target nodes after diffusion.

[0060] The following examples will provide a detailed description of the four stages mentioned above:

[0061] First, we introduce Step A: the graph construction phase. In an exemplary embodiment, Figure 3 This is a schematic flowchart illustrating the method for constructing a graph relationship network provided in the embodiments of this specification. (See reference...) Figure 3 The embodiment shown in the figure includes S310-S320.

[0062] In S310, a message passing mechanism based on the BSP computing model is used to store the private messages corresponding to each node.

[0063] The nodes, edges connecting different nodes, and self-loop edges constitute the graph relationship network described above. In the embodiments of this specification, users and media can be used as nodes. The media can be devices, documents (ID cards, driver's licenses, etc.), or identification numbers (mobile phone numbers), etc.

[0064] Since each node in a graph relational network may have relational edges and be associated with multiple other nodes, a fixed private message for each node is stored using the Pregel message mechanism to facilitate information transmission.

[0065] In S320, the relationship edges between nodes are determined based on the relationship data between nodes, and the self-loop edges of nodes are determined based on the support data of nodes, thus obtaining the graph relationship network.

[0066] For example, if user s uses mobile phone a to make a payment x, then user s and mobile phone a can be represented as node 1 and node 2, respectively. Since there is a relationship between user s and mobile phone a, the directed connection between node 1 and node 2 can be determined, that is, the relationship edge between node 1 and node 2 can be determined. In this embodiment, user a's payment behavior x can be represented as a self-loop edge of node 1.

[0067] in, Figure 4 This is a schematic diagram of the graph relationship network provided in the embodiments of this specification. (See reference) Figure 4 The graph relationship network shown does not include self-loop edges. Node P... a Node P b Node P c Node P d Node P e Node P h Node P j and node P k Representing users, node E ab Node E ac Node E ad Node E ae Node E hj and node E hk Each node represents a device, and the relationships between them are shown in the image below. Figure 4 As shown.

[0068] After constructing the graph relationship network through Step A above, Step B: Graph Traversal Stage is executed. In Step B, multiple subgraphs are first identified in the entire graph relationship network. Then, for each subgraph, the starting node (target node in the embodiment of this specification) in the subgraph is traversed to record the key information of the relationship edges traversed by the private message from the target node to the reaching node (recorded as the attributes of the relationship edges, including the actual association time and the actual association number).

[0069] First, we introduce the process of identifying subgraphs from the constructed graph relationship network. During Pregel-based data processing, multiple machines in the cluster are selected to execute the data processing tasks. Each machine runs a copy of the program. One machine in the cluster is selected as the Master, and the others act as Workers. Then, the Master divides the graph relationship network into multiple subgraphs and assigns them to multiple Workers. Each Worker receives one or more subgraphs. Each Worker is responsible for maintaining the automatic state (addition and deletion of nodes and edges) assigned to it, executing the `Compute()` function on the vertices in its assigned partition, sending messages outwards, and managing received messages.

[0070] In this embodiment of the specification, a subgraph is constructed by including all devices associated with the current node and all other groups of people associated with those devices. For example, Figure 5a and Figure 5b According to Figure 4 The diagram shows a subgraph determined by the relationship network shown. (Reference) Figure 5a The subgraph shown has node P as the current node. h All devices associated with it are: Node E hj Node E hk and node E ab And all other groups of people associated with these devices, including: Node P j Node P k and node P b . refer to Figure 5b The subgraph shown has node P as the current node. a All devices associated with it are: Node E ab Node E ac Node E ad and node E ae And all other groups of people associated with these devices, including: Node P b Node P c Node P d and node P e .

[0071] For each subgraph, for ease of understanding, nodes are distinguished into nodes corresponding to the risk initiator and nodes corresponding to the risk trigger (group). The node corresponding to the risk initiator is the target node for calculating the risk contribution value. In this embodiment, the node corresponding to the risk trigger (group) is denoted as the reaching node. (See reference...) Figure 5a The node corresponding to the risk initiator is node P. h The node corresponding to the risk trigger person (group) is: node P. j Node P k and node P b . refer to Figure 5b The node corresponding to the risk initiator is node P. a The node corresponding to the risk trigger person (group) is: node P. b Node P c Node P d and node P e .

[0072] In this embodiment of the specification, after determining the subgraph from the constructed graph relationship network, each subgraph is traversed starting from the node of the risk initiator, where the traversal process corresponds to the propagation path. In this embodiment of the specification, the path by which the private message of the target node is propagated from the target node to the aforementioned reachable nodes is denoted as the propagation path. It can be understood that the propagation path consists of at least one relation edge. (See reference...) Figure 5a Node P h The private message of the target node is sent by node P. h (Target Node) propagates to the aforementioned node P j The propagation path (reaching the node): Node P h →Node E hj →Node P j Node P h The private message of the target node is sent by node P. h (Target Node) propagates to the aforementioned node P k The propagation path (reaching the node): Node P h →Node E hk →Node P k Node P h The private message of the target node is sent by node P. h (Target Node) propagates to the aforementioned node P b The propagation path (reaching the node): Node P h →Node E ab →Node P b ;visible, Figure 5a The subgraph shown includes 3 propagation paths, and similarly... Figure 5b The subgraph shown includes 6 propagation paths.

[0073] In the embodiments of this specification, for each subgraph, not all private messages from target nodes will be successfully sent to the nodes of each risk triggerer (group). Therefore, it is necessary to determine which relationship edges are trustworthy based on pre-set filtering conditions, while corresponding untrustworthy edges will be filtered out. This ensures the accuracy of risk contribution values ​​while improving data processing efficiency. Specifically, the embodiments of this specification compare the above filtering conditions with the attributes of the relationship edges, and determine which relationship edges are trustworthy and which are untrustworthy based on the comparison results. It can be understood that private messages can reach the reachable nodes from the target nodes through trustworthy relationship edges, and these trustworthy relationship edges are also called "reaching edges". Further, in subsequent embodiments of this specification, the self-loop edges of the reaching nodes will be subject to trigger-based filtering based on the attributes corresponding to the reaching edges.

[0074] Furthermore, embodiments of this specification will describe the attributes of relation edges, and embodiments comparing the attributes of relation edges with preset filtering conditions. Exemplary References Figure 5b Propagation path: Node P a →Node E ab →Node P b It contains two relation edges: L ab 1 and L ab 2. Each relation edge has an attribute that expresses the relationship between the two nodes it connects, including the actual association time and / or the actual number of associations between the two nodes. For example, for the relation edge L mentioned above... ab 1. The corresponding attribute represents node P. a With node E ab The actual time and / or number of actual associations between them.

[0075] For example, the association time in the attribute can refer to the time point when one node in a relationship edge triggers another node, such as node P. a The corresponding user a passes through node E ab If the corresponding device s logs in to a certain application, then the relation edge L... ab The attribute of 1 includes the time point at which user a triggers device s to perform the application login behavior.

[0076] For example, the association time in the attribute can also refer to the duration during which one node in a relation edge performs a certain action through another node, such as node P. a The corresponding user a passes through node E ab If the corresponding device s performs product evaluation, then the relation edge L ab The attribute of 1 includes the duration of time that user a performs product review behavior through device s.

[0077] For example, the association count in the attribute refers to the number of interactions between connection points within a unit, such as node P. a The corresponding user a passes through node E ab If the corresponding device s logs into a certain application 200 times within a unit of time (10 minutes), then the relation edge L... ab The attribute of 1 includes the number of times user a logs into an application through device s within a unit of time.

[0078] In an exemplary embodiment, Figure 6 This is a schematic flowchart illustrating the relation edge filtering method provided in the embodiments of this specification. (Reference) Figure 6 In S610, for any relation edge in the subgraph, the attribute corresponding to the relation edge is compared with the preset filtering conditions. In S620, if the attribute corresponding to the relation edge does not meet the above filtering conditions, the propagation path containing the relation edge is filtered out.

[0079] The preset filtering conditions include constraint association duration and preset association count. In the embodiments of this specification, the association time in the attribute of the relation edge is denoted as "actual association time," and the association count in the attribute is denoted as "time association count," and these are used to compare with the constraint association duration and preset association count included in the filtering conditions, respectively. Specifically, for the target relation edge (such as the relation edge L mentioned above) ab 1) In the context of the actual association time in its attributes representing the duration for one node in a relation edge to perform a certain action through another node, determine whether the actual association time falls within the aforementioned constraint association time range. If the aforementioned relation edge L... ab If the actual association time in attribute 1 is 5 minutes, and the constraint association time range is [1 minute, 3 minutes], then the above relationship edge L can be determined. ab The actual association time in attribute 1 does not satisfy the constraint association time. For the target relation edge (such as relation edge L mentioned above)... ab 1) If the actual number of associations in its attributes is 50, and the constraint association number range is [1, 3], then the above relation edge L can be determined. ab The actual number of associations in attribute 1 does not meet the constraint association time.

[0080] In this embodiment of the specification, for any relation edge, if any one of its attributes does not meet the preset filtering condition, the relation edge is considered untrustworthy, the message cannot be transmitted through the target edge, and the propagation path of the relation edge is not actually used for message transmission. Therefore, if the attribute corresponding to the relation edge does not meet the above filtering condition, the path containing the relation edge is filtered out. Exemplary Reference Figure 7 By comparison, the relation edge L can be determined. ae 2 represents an untrusted relation edge, even if the relation edge L...ae 1 is a trusted edge, but because it comes from node P a The message cannot pass through node E ae Passed to node P e Then the relation edge L ae The propagation path of node P is as follows: a →Node E ae →Node P e This needs to be filtered out. Continue to refer to... Figure 7 By comparison, the relation edge L can be determined. ad 1 represents an untrusted relation edge, even if the relation edge L... ad 2 and Lad2' are trusted edges, but because they come from node P a The message cannot be transmitted through node E ad Passed to node P d and node P e Then the relation edge L ae The propagation path of node P is as follows: a →Node E ad →Node P e And the propagation path: node P a →Node E ad →Node P d They need to be screened out.

[0081] Through such Figure 6 The illustrated embodiment, after filtering out the propagation path containing untrusted relationship edges, ensures that the target node's private message is successfully sent to the node of the relevant risk trigger (group) via the remaining propagation path (e.g., ...). Figure 7 Middle node P b and node P c) This ensures the accuracy of risk contribution values ​​while also improving the timeliness of data processing.

[0082] In an exemplary embodiment, Figure 8 This is a schematic flowchart illustrating a data processing method provided in another embodiment of this specification. The embodiment shown in the figure is... Figure 1 It was based on that. The difference lies in... Figure 8 In the illustrated embodiment, after executing S110, S810 and S820 are executed. S810 and S820 provide the attributes corresponding to each propagation path for comparison with preset constraints in S120. (See reference...) Figure 8 :

[0083] In S810, for the relation edges contained in the propagation path, the actual association time and / or the actual number of associations between the two nodes connected by the relation edge are determined as the attribute corresponding to the relation edge. And in S820, based on the attribute corresponding to the relation edges contained in the propagation path, the attribute corresponding to the propagation path is determined.

[0084] It is understandable that each propagation path contains at least one relation edge, such as a reference. Figure 5b Propagation path: Node P a →Node E ab →Node P b It contains two relation edges: L ab 1 and L ab 2. Each relation edge corresponds to an attribute that expresses the relationship between the two nodes it connects. In the embodiments of this specification, the attributes of the propagation path are determined based on the attributes of the relation edges contained in the propagation path. Taking the above example, based on the relation edge L... ab 1. Its corresponding attributes and relation edges L ab 2. Its corresponding attributes determine the propagation path: Node P a →Node E ab →Node P b The corresponding attributes. Specifically:

[0085] In one exemplary embodiment, for a propagation path containing two nodes:

[0086] If a propagation path includes a target node and a destination node in sequence (e.g., ... Figure 5a Propagation path: Node P h →Node P k The actual association time and / or actual association count between the target node and the reach node are determined as the attribute corresponding to the relation edge connecting the two nodes; furthermore, since the propagation path contains a relation edge, the attribute corresponding to the relation edge is determined as the attribute corresponding to the propagation path.

[0087] In another exemplary embodiment, for a propagation path with more than two nodes:

[0088] This specification provides examples of a scenario involving three nodes. If a propagation path sequentially includes a target node, an intermediate node, and a destination node, then it can be achieved through... Figure 7 The illustrated embodiment determines the properties of the propagation path. (See reference...) Figure 9 The embodiment shown in the figure includes: S8102 and S8104 as one implementation of S810, and S8202 and S8204 as one implementation of S820.

[0089] In S8102, the actual association time and / or the actual number of associations between the target node and the intermediate node are determined as the attributes corresponding to the first relation edge. And in S8104, the actual association time and / or the actual number of associations between the intermediate node and the reached node are determined as the attributes corresponding to the second relation edge.

[0090] For example, with Figure 7 Propagation path: Node P a →Node E ab →Node P b For example, this propagation path includes the target node P. a Intermediate node E ab and reaching node P b , where the target node P a With intermediate node E ab The actual association time and / or the actual number of associations between them is the first relation edge L. ab The attribute corresponding to 1 is the node E. ab With the reach node P b The actual association time and / or the actual number of associations between them is the second relation edge L. ab The attribute corresponding to 2.

[0091] Furthermore, by traversing the subgraph, attributes of the reachable edges can be obtained. For each reachable node, this can represent the attributes of the reachable edges contained in all the propagation paths traversed to reach it. For example, for reachable node P... b It receives from node P a and node P h Private messages, from target node P a The private message is propagated through path 1 (node ​​P) a →Node E ab →Node P b ) and propagation path 2 (node ​​P) a →Node E ac →Node P b Reaching the aforementioned reach node P b From target node P h The private message is propagated through path 3 (node ​​P) h →Node E hb →Node P b Reaching the aforementioned reach node P b Reaching the aforementioned node P b The attributes of the reachable edges contained in the three propagation paths mentioned above can be represented as follows:

[0092]

[0093] The row number of this matrix indicates the node P that is reached. b The number of messages received (which also corresponds to the number of propagation paths reaching the node) is recorded in the matrix above, representing the private messages from the target node P. a P h To reach node P b The key information of the relationships traversed during the process (attributes of the reached edges, including actual association time and actual number of associations), specifically: tA ob1 andtA qb1 This indicates that in the above propagation path 1, the reach edge (node ​​P) is... a →Node E ab , denoted as a first-degree relation edge, where tA ob1 This indicates the actual association time of the first relation edge in propagation path 1, tA. qb1 tA represents the actual number of associations of the first relation edge in propagation path 1. ob2 andtA qb2 This indicates that in the above propagation path 1, the reach edge (node ​​E) is... ab →Node P b , denoted as the attribute of a second-degree relation edge, where tA ob2 Indicates the actual association time of the second relation edge in propagation path 1, tA qb2 This indicates the actual number of associations of the second relation edge in propagation path 1;

[0094] tA oc1 andtA qc1 This indicates that in the above propagation path 2, the reach edge (node ​​P) is... a →Node E ac , denoted as a first-degree relation edge, where tA oc1 Indicates the actual association time of the first relation edge in propagation path 2, tA qc1 tA represents the actual number of associations of the first relation edge in propagation path 2. oc2 andtA qc2 This indicates that in the above propagation path 2, the reach edge (node ​​E) is... ac →Node P b , denoted as the attribute of a second-degree relation edge, where tA oc2 This indicates the actual association time of the second relation edge in propagation path 2, tA. qc2 This indicates the actual number of associations of the second relation edge in propagation path 2;

[0095] tB ob1 and tB qb1 This indicates that in the above propagation path 3, the reach edge (node ​​P) is... h →Node E ab , denoted as a first-degree relation edge, where tBob1 This indicates the actual association time (tB) of the first relation edge in propagation path 3. qb1 tB represents the actual number of associations of the first relation edge in propagation path 3. ob2 and tB qb2 This indicates that in the above propagation path 3, the reach edge (node ​​E) is... ab →Node P b , denoted as a second-degree relation edge, has attributes where tB ob2 This indicates the actual association time of the second relation edge in propagation path 3, tA. qb2 This indicates the actual number of associations of the second relation edge in propagation path 3.

[0096] The following examples will be described. Figure 2 Step C, as shown, pertains to the calculation of reach nodes. Similar to the previous embodiment, the matrix corresponding to each reach node can be determined. In a credit scenario, for each reach node, its disbursement behavior is represented by the self-loop edges of that node, as shown in the reference... Figure 10 An example is shown regarding node P b Two self-loop edges. An exemplary self-loop edge reaching a node can be represented as X. p =[x p1 x p2 x p3 x p4 x p5 ...]. In this embodiment, the aforementioned matrix information of the reached node is used to filter the self-loop edges of the reached node when they are triggered:

[0097] Fliter(X p When β1≤O≤β2, Q≥θ

[0098] Where O represents the actual association time in the attribute corresponding to the propagation path, Q represents the actual association number in the attribute corresponding to the propagation path, β1 and β2 are the preset association time range in the preset constraints, and θ is the preset association number in the preset constraints.

[0099] As can be seen, after determining the attributes of each relation edge through S810, the attributes corresponding to the propagation path are further determined based on the attributes corresponding to the first relation edge and the attributes corresponding to the second relation edge. Then, the attributes corresponding to the propagation path are compared with the preset constraints. For details, please refer to... Figure 9In S8202, the actual association time in the attributes of the propagation path is determined based on the actual association time in the attributes corresponding to the first relation edge and the actual association time in the attributes corresponding to the second relation edge; in S8204, the actual association count in the attributes of the propagation path is determined based on the actual association count in the attributes of the first relation edge and the actual association count in the attributes of the second relation edge.

[0100] For example, regarding the actual association time in the attributes corresponding to a propagation path, if the association time represents duration, the maximum / minimum actual association time of the relation edge attributes in the propagation path can be determined, or the actual association duration of each relation edge can be summed to determine the actual association time. Specifically, any of the above methods can be used to determine the actual association time in the attributes corresponding to the propagation path, depending on the actual situation. For example, the first relation edge L... ab The actual association time in the attribute corresponding to 1 is 5 minutes, and the second relation edge L ab If the actual association time of the attribute corresponding to 2 is 3 minutes, then the time association time of the attribute corresponding to the propagation path formed by the above two relation edges can be 5 minutes, 3 minutes or 8 minutes.

[0101] For example, regarding the actual association time in the attribute corresponding to the propagation path, when the association time represents a point in time, the actual association duration of each relation edge (reaching edge) in the propagation path can be used as the actual association time in the attribute corresponding to the propagation path. Alternatively, the maximum actual association time of the relation edge attribute in the propagation path can be determined as the actual association time in the attribute corresponding to the propagation path. Specifically, any of the above methods can be used to determine the actual association time in the attribute corresponding to the propagation path based on the actual situation. For example, the first relation edge L... ab The actual association time in the attribute corresponding to 1 is: November 12, 2022, 13:00, second relation edge L. ab The actual associated time in the attribute corresponding to 2 is 13:01 on November 12, 2022. Therefore, the time associated time of the attribute corresponding to the propagation path formed by the above two relation edges can be 13:00 on November 12, 2022, 13:01 on November 12, 2022, or 13:00 on November 12, 2022 and 13:01 on November 12, 2022.

[0102] For example, the actual number of associations in the attributes corresponding to a propagation path can be determined by summing the maximum / minimum actual number of associations for each relation edge in the propagation path. Specifically, any of the above methods can be used to determine the actual number of associations in the attributes corresponding to the propagation path, depending on the actual situation. For example, the first relation edge L... abThe attribute corresponding to 1 is actually associated once, and the second relation edge L ab If the actual association frequency of the attribute corresponding to 2 is 3 times, then the time association frequency of the attribute corresponding to the propagation path formed by the above two relation edges can be 1 time, 3 times, or 4 times.

[0103] pass Figure 8 The solution provided in the illustrated embodiment can determine the attributes (including actual association time O and / or actual association count Q) of each propagation path for each traversable node in each subgraph.

[0104] Next, we will introduce... Figure 2 Step D, as shown, pertains to the target node calculation phase. Specifically, it involves calculating the risk value of the target node. First, the nodes that received the private message from the target node are identified. Then, the self-loop edges of these nodes are filtered to identify the target behavior associated with that private message. Finally, node P is determined based on the target behavior corresponding to the filtered self-loop edges of the relevant nodes. a The risk value.

[0105] Regarding the aforementioned trigger-based filtering process: it is achieved by comparing the attributes corresponding to the propagation path with preset constraints. Specifically: by comparing the actual association time and constraint association time in the attributes corresponding to the propagation path, a first target behavior group that satisfies the aforementioned constraint association time is determined; by comparing the actual association count and constraint association count in the attributes corresponding to the propagation path, a second target behavior group that satisfies the aforementioned constraint association count is determined; the intersection of the first target behavior group and the second target behavior group is determined as the filtered partial self-loop edge, wherein the behavior represented by the filtered partial self-loop edge is the target behavior associated with the aforementioned private message.

[0106] In the embodiments of this specification, the target node P is used as an example. a The calculation of risk value will be introduced as an example: (Reference) Figure 4 and Figure 7 In order to compute node P a The risk value requires determining that the node receiving the private message from node Pa is node P. b and node P c Next, we will work on node P. b and node P c The self-loop edges are used for filtering, and finally, based on node P... b and node P c The filtered part determines the node P by the self-loop edge. a The risk value.

[0107] Specifically, for the reached node P bOf its three propagation paths, the one used to transmit data from node P is... a There are two propagation paths, namely propagation path 1 and propagation path 2. Assume that the actual association time O in the attributes corresponding to the two propagation paths is... b It can be represented as: [O b1 O b2 The actual association time Q in the attributes corresponding to the two propagation paths mentioned above. b It can be represented as: [Q b1 Q b2 Additionally, its self-loop edge is represented as X. pb =[x pb1 x pb2 x pb3 x pb4 ].

[0108] Then utilize the reach node P b The above matrix information is used to filter the self-loop edges of the reached node when triggering:

[0109] Fliter(X pb ), when β1≤O b ≤β2, Q b ≥θ

[0110] For example, compare the actual association time O in the attributes corresponding to propagation path 1. b1 Given the constraint association time [β1, β2], determine the first target behavior group that satisfies the constraint association time, for example, [x pb1 x pb2 x pb4 ]; Compare the actual number of associations Q in the attributes corresponding to propagation path 1. b1 Given the number of constraint associations θ, determine the second target behavior group [x] that satisfies the number of constraint associations. pb1 x pb2 ]; the first target behavior group [x pb1 x pb2 x pb4 ] and the second target behavior group [x pb1 x pb2 The intersection of [x] is used to determine the selected portion of the self-loop edge [x]. pb1 x pb2 ], where the selected portion is from the loop edge [x pb1 x pb2 The action represented by ] is related to the above node P. a The behavior associated with private messages.

[0111] It also compares the actual association time O in the attributes corresponding to propagation path 2. b2Given the constraint association time [β1, β2], determine the first target behavior group that satisfies the constraint association time, for example, [x pb1 x pb2 x pb3 ]; Compare the actual number of associations Q in the attributes corresponding to propagation path 2. b2 Given the number of constraint associations θ, determine the second target behavior group [x] that satisfies the number of constraint associations. pb1 x pb2 ]; the first target behavior group [x pb1 x pb2 x pb3 ] and the second target behavior group [x pb1 x pb2 The intersection of [x] is used to determine the selected portion of the self-loop edge [x]. pb1 x pb2 ], where the selected portion is from the loop edge [x pb1 x pb2 The action represented by ] is related to the above node P. a The behavior associated with private messages.

[0112] In summary, for the reached node P b On the self-loop edge X pb =[x pb1 x pb2 x pb3 x pb4 ] Filter out x pb1 x pb2 This means x pb1 x pb2 The corresponding spending behavior meets the conditions.

[0113] For the reached node P c There is only one propagation path (node ​​P) a →Node E ac →Node P c Let this be denoted as propagation path 4). If the actual association time in the attribute corresponding to propagation path 4 can be represented as O... c The actual association time in the attribute corresponding to the above propagation path 4 can be represented as Q. c Additionally, its self-loop edge is represented as X. pc =[x pc1 x pc2 x pc3 ].

[0114] Then utilize the reach node P c The above matrix information is used to filter the self-loop edges of the reached node when triggering:

[0115] Fliter(X pc), when β1≤O c ≤β2, Q c ≥θ

[0116] For example, on the self-loop edge X pc =[x pc1 x pc2 x pc3 ] Filter out x pc2 This means x pc2 The corresponding spending behavior meets the conditions.

[0117] Furthermore, based on the self-loop edge x that satisfies the conditions pb1 x pb2 and x pc2 The risk value of node Pa is determined by summing the target behaviors corresponding to the selected self-loop edges. In an exemplary embodiment, the risk value of the target node is obtained by summing the target behaviors corresponding to the selected self-loop edges; for example, the payment amount of the payment behavior, assuming the self-loop edge x pb1 x pb2 and x pc2 The corresponding payment amounts are z1, z2, and z3 yuan respectively, so node P can be determined. a The risk value can be expressed as z1 + z2 + z3. Of course, it can also be converted into other units of measurement based on the amount, and there is no restriction on this.

[0118] In another exemplary embodiment, the weights of the target behaviors corresponding to the selected self-loop edges are determined, and the products of the target behaviors and their corresponding weights are summed to obtain the risk value corresponding to the target node. For example, the payment amount of the payment behavior, assuming the self-loop edge x pb1 x pb2 and x pc2 The corresponding payment amounts are z1, z2, and z3 yuan respectively, and the self-loop edge x is determined. pb1 x pb2 and x pc2 If the corresponding weights are α1, α2, and α,3 respectively, then node P can be determined. a The risk value can be expressed as α1×z1+α2×z2+α3×z3. Of course, it can also be converted to other units of measurement based on the amount; there are no restrictions on this.

[0119] The aforementioned risk value identifies the risk value (also called risk contribution) posed by all reachable nodes after the target node spreads. Based on the assessed and calculated risk contribution, individual and overall risk profiles can be created, enabling early warning and risk interception. For example, after determining the risk value corresponding to the target node, if the risk value of the target node is greater than a preset threshold, a warning message about the target node is generated.

[0120] The solution provided in the embodiments of this specification determines the propagation path of the private messages stored by a target node in a graph relation network. These private messages are sent using a message passing mechanism based on a synchronous parallel BSP computing model, thereby enabling parallel processing of each node in the graph relation network and improving data processing efficiency. Specifically, by comparing the attributes corresponding to the propagation path with preset constraints, some self-loop edges satisfying the constraints are identified from all self-loop edges reaching the node in the propagation path. Further, based on the target behaviors corresponding to these self-loop edges, the risk value corresponding to the target node can be determined. Therefore, in credit risk control scenarios, the embodiments of this specification can efficiently and accurately calculate the contribution of personal fraud risk, thereby effectively reducing financial risk. For example, it effectively curbs phenomena such as multiple identity impersonation, collusion, cash-out schemes, coupon fraud, order brushing, fake reviews, and shipping fraud.

[0121] It should be noted that the above figures are merely illustrative of the processes included in the methods according to exemplary embodiments of this specification, and are not intended to be limiting. It is readily understood that the processes shown in the above figures do not indicate or limit the temporal order of these processes. Furthermore, it is readily understood that these processes may, for example, be executed synchronously or asynchronously in multiple modules.

[0122] The following are embodiments of the apparatus described in this specification, which can be used to execute the embodiments of the methods described in this specification. For details not disclosed in the apparatus embodiments of this specification, please refer to the embodiments of the methods described in this specification.

[0123] in, Figure 11 This is a schematic diagram of the data processing apparatus provided in the embodiments of this specification. Please refer to [link / reference]. Figure 11 The data processing device shown in the figure can be implemented as all or part of an electronic device through software, hardware, or a combination of both, or it can be integrated into an electronic device as an independent module.

[0124] In the embodiments described in this specification, the data processing device 1100 includes: a path determination module 1110, a self-loop edge filtering module 1120, and a risk value determination module 1130.

[0125] The path determination module 1110 is used to determine the propagation path of the private message stored by the target node in the graph relation network, wherein the private message is sent by the message passing mechanism based on the synchronous parallel BSP computing model, and the propagation path includes the reaching node; the self-loop edge filtering module 1120 is used to determine some self-loop edges among all self-loop edges of the reaching node by comparing the attributes corresponding to the propagation path and the preset constraints; and the risk value determination module 1130 is used to determine the risk value corresponding to the target node according to the target behavior corresponding to the aforementioned partial self-loop edges.

[0126] In an exemplary embodiment, based on the above scheme, the method further includes: a message passing mechanism based on the BSP computing model to save the private messages corresponding to each node; determining the relationship edges between nodes based on the relationship data between nodes, and determining their self-loop edges based on the usage data of nodes, to obtain the above graph relationship network; wherein the above nodes include: users and media, and the above media includes one or more of devices, certificates and identification numbers.

[0127] In an exemplary embodiment, based on the above scheme, the method further includes: determining a subgraph in the graph relationship network, wherein the subgraph includes: an initiating node corresponding to a risk initiator and an access node corresponding to a risk triggerer, and the access node includes one or more.

[0128] For a target node in a graph relation network, determining the propagation path of the private message stored in the target node includes: determining the starting node in the subgraph as the target node; and determining the path of the private message propagating from the target node to the reaching node as the propagation path.

[0129] In an exemplary embodiment, based on the above scheme, the method further includes: for the relation edges included in the propagation path, determining the actual association time and / or the actual association count between the two nodes connected by the relation edges as the attribute corresponding to the relation edges; and determining the attribute corresponding to the propagation path based on the attribute corresponding to the relation edges included in the propagation path.

[0130] In an exemplary embodiment, based on the above scheme, determining the actual association time and / or actual association count between the two nodes connected by the relationship edge in the propagation path as the attribute corresponding to the relationship edge includes: when the propagation path sequentially includes the target node and the reach node, determining the actual association time and / or actual association count between the target node and the reach node as the attribute corresponding to the relationship edge.

[0131] The above determination of the attributes corresponding to the propagation path based on the attributes corresponding to the relation edges contained in the propagation path includes: determining the attributes corresponding to the relation edges as the attributes corresponding to the propagation path.

[0132] In an exemplary embodiment, based on the above scheme, determining the actual association time and / or actual association count between the two nodes connected by the relationship edge included in the propagation path as the attribute corresponding to the relationship edge includes: when the propagation path sequentially includes the target node, an intermediate node, and the reach node, determining the actual association time and / or actual association count between the target node and the intermediate node as the attribute corresponding to the first relationship edge; and determining the actual association time and / or actual association count between the intermediate node and the reach node as the attribute corresponding to the second relationship edge.

[0133] The above determination of the attributes corresponding to the propagation path based on the attributes corresponding to the relation edges contained in the propagation path includes: determining the attributes corresponding to the propagation path based on the attributes corresponding to the first relation edge and the attributes corresponding to the second relation edge.

[0134] In an exemplary embodiment, based on the above scheme, determining the attribute corresponding to the propagation path according to the attribute corresponding to the first relation edge and the attribute corresponding to the second relation edge includes: determining the actual association time in the attribute of the propagation path according to the actual association time in the attribute corresponding to the first relation edge and the actual association time in the attribute corresponding to the second relation edge; and / or, determining the actual association count in the attribute of the propagation path according to the actual association count in the attribute of the first relation edge and the actual association count in the attribute of the second relation edge.

[0135] In an exemplary embodiment, based on the above scheme, determining the propagation path of the private message stored by the target node in the graph relation network includes: for any relation edge in the subgraph, comparing the attribute corresponding to the relation edge with a preset filtering condition; and, if the attribute corresponding to the relation edge does not meet the filtering condition, filtering out the propagation path where the relation edge is located, wherein the remaining propagation path after filtering is the propagation path of the private message stored by the target node, and comparing the attribute corresponding to it with the above constraint condition.

[0136] In an exemplary embodiment, based on the above scheme, the attributes corresponding to the relation edge include the actual association time and the actual association count, and the filtering conditions include the filtering association time and the filtering association count. The comparison of the attributes corresponding to the relation edge with the preset filtering conditions includes: for a target relation edge, comparing the actual association time and the filtering association time in the attributes corresponding to the target relation edge, and comparing the actual association count and the filtering association count in the attributes corresponding to the target relation edge. Wherein, if the actual association time does not meet the filtering association time, and / or the actual association count does not meet the filtering association count, the propagation path where the target relation edge is located will be filtered out.

[0137] In an exemplary embodiment, based on the above scheme, the attributes corresponding to the propagation path include actual association time and actual association count, and the constraints include constraint association time and constraint association count. The method of determining a partial self-loop edge among all self-loop edges of the reached node by comparing the attributes corresponding to the propagation path with the preset constraints includes: comparing the actual association time in the attributes corresponding to the propagation path with the constraint association time to determine a first target behavior group that satisfies the constraint association time; comparing the actual association count in the attributes corresponding to the propagation path with the constraint association count to determine a second target behavior group that satisfies the constraint association count; and determining the intersection of the first target behavior group and the second target behavior group as the partial self-loop edge, wherein the behavior represented by the partial self-loop edge is a target behavior associated with the private message.

[0138] In an exemplary embodiment, based on the above scheme, determining the risk value corresponding to the target node according to the target behaviors corresponding to the aforementioned partial self-loop edges includes: summing the target behaviors corresponding to the aforementioned partial self-loop edges to obtain the risk value corresponding to the target node; or, determining the weights of the target behaviors corresponding to the aforementioned partial self-loop edges, multiplying the target behaviors by their corresponding weights, and then summing the products to obtain the risk value corresponding to the target node.

[0139] In an exemplary embodiment, based on the above scheme, after determining the risk value corresponding to the target node, the method further includes: generating warning information about the target node when it is determined that the risk value corresponding to the target node is greater than a preset threshold.

[0140] It should be noted that the decision-making device provided in the above embodiments is only illustrated by the division of the above functional modules when executing the decision-making method. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above.

[0141] Furthermore, the decision-making device and decision-making method embodiments provided in the above embodiments belong to the same concept. Therefore, for details not disclosed in the device embodiments of this specification, please refer to the above-described embodiments of the decision-making method, which will not be repeated here.

[0142] This specification also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of any of the methods described in the foregoing embodiments. The computer-readable storage medium may include, but is not limited to, any type of disk, including floppy disks, optical disks, DVDs (Digital Video Discs), CD-ROMs (Compact Disc Read-Only Memory), microdrives, magneto-optical disks, ROMs (Read-Only Memory), RAMs (Random Access Memory), EPROMs (Erasable Programmable Read-Only Memory), EEPROMs (Electrically Erasable Programmable Read-Only Memory), DRAMs (Dynamic Random Access Memory), VRAMs (Video Random Access Memory), flash memory devices, magnetic cards or optical cards, nanosystems (including molecular memory ICs), or any type of media or device suitable for storing instructions and / or data.

[0143] Figure 12 This schematic diagram illustrates the structure of an electronic device according to an exemplary embodiment of this specification. Please refer to... Figure 12 As shown, the electronic device 1200 includes a processor 1201 and a memory 1202.

[0144] In this embodiment, processor 1201 is the control center of the computer system and can be a processor of a physical machine or a processor of a virtual machine. Processor 1201 may include one or more processing cores, such as a 4-core processor or an 8-core processor. Processor 1201 may be implemented using at least one hardware form selected from Digital Signal Processing (DSP), Field-Programmable Gate Array (FPGA), and Programmable Logic Array (PLA). Processor 1201 may also include a main processor and a coprocessor; the main processor is used to process data in the wake-up state, and the coprocessor is a low-power processor used to process data in the standby state.

[0145] In the embodiments of this specification, the processor 1201 is specifically used to: for a target node in a graph relation network, determine the propagation path of the private message stored in the target node, wherein the private message is sent by a message passing mechanism based on a synchronous parallel BSP computing model, and the propagation path includes reaching nodes; by comparing the attributes corresponding to the propagation path and preset constraints, determine some self-loop edges among all self-loop edges of the reaching nodes; and determine the risk value corresponding to the target node based on the target behavior corresponding to the aforementioned partial self-loop edges.

[0146] Furthermore, the processor 1201 is specifically used for: a message passing mechanism based on the BSP computing model to store private messages corresponding to each node; and, based on the relationship data between nodes, to determine the relationship edges between nodes, and based on the support data of nodes, to determine their self-loop edges, thereby obtaining the graph relationship network; wherein the nodes include: users and media, and the media includes one or more of devices, certificates, and identification numbers.

[0147] Furthermore, the processor 1201 is specifically configured to: determine a subgraph in the graph relationship network, wherein the subgraph includes: an initiating node corresponding to a risk initiator and an access node corresponding to a risk triggerer, wherein the access node includes one or more.

[0148] For a target node in a graph relation network, determining the propagation path of the private message stored in the target node includes: determining the starting node in the subgraph as the target node; and determining the path of the private message propagating from the target node to the reaching node as the propagation path.

[0149] Furthermore, the processor 1201 is specifically configured to: determine the actual association time and / or actual association count between the two nodes connected by the relation edge in the propagation path as the attribute corresponding to the relation edge; and determine the attribute corresponding to the propagation path based on the attribute corresponding to the relation edge in the propagation path.

[0150] Furthermore, for the relation edges included in the above propagation path, determining the actual association time and / or the actual association count between the two nodes connected by the relation edge as the attribute corresponding to the relation edge includes: when the above propagation path sequentially includes the target node and the reach node, determining the actual association time and / or the actual association count between the target node and the reach node as the attribute corresponding to the relation edge.

[0151] The above determination of the attributes corresponding to the propagation path based on the attributes corresponding to the relation edges contained in the propagation path includes: determining the attributes corresponding to the relation edges as the attributes corresponding to the propagation path.

[0152] Furthermore, regarding the relation edges included in the aforementioned propagation path, determining the actual association time and / or actual association count between the two nodes connected by the relation edge as the attribute corresponding to the relation edge includes: when the aforementioned propagation path sequentially includes the aforementioned target node, an intermediate node, and the aforementioned reach node, determining the actual association time and / or actual association count between the aforementioned target node and the aforementioned intermediate node as the attribute corresponding to the first relation edge; and determining the actual association time and / or actual association count between the aforementioned intermediate node and the aforementioned reach node as the attribute corresponding to the second relation edge;

[0153] The above determination of the attributes corresponding to the propagation path based on the attributes corresponding to the relation edges contained in the propagation path includes: determining the attributes corresponding to the propagation path based on the attributes corresponding to the first relation edge and the attributes corresponding to the second relation edge.

[0154] Furthermore, determining the attribute corresponding to the propagation path based on the attribute corresponding to the first relation edge and the attribute corresponding to the second relation edge includes: determining the actual association time in the attribute of the propagation path based on the actual association time in the attribute corresponding to the first relation edge and the actual association time in the attribute corresponding to the second relation edge; and / or, determining the actual association count in the attribute of the propagation path based on the actual association count in the attribute of the first relation edge and the actual association count in the attribute of the second relation edge.

[0155] Furthermore, the above-mentioned determination of the propagation path of the private message stored by the target node in the graph relation network includes: for any relation edge in the subgraph, comparing the attribute corresponding to the relation edge with the preset filtering conditions; and, if the attribute corresponding to the relation edge does not meet the filtering conditions, filtering out the propagation path where the relation edge is located, wherein the remaining propagation path after filtering is the propagation path of the private message stored by the target node, and the attribute corresponding to it is compared with the above-mentioned constraints.

[0156] Furthermore, the attributes corresponding to the aforementioned relation edges include the actual association time and the actual number of associations, and the aforementioned filtering conditions include filtering association time and filtering the number of associations. The comparison of the attributes corresponding to the aforementioned relation edges with the preset filtering conditions includes: for a target relation edge, comparing the actual association time and the filtered association time in the attributes corresponding to the target relation edge, and comparing the actual number of associations and the filtered association time in the attributes corresponding to the target relation edge; wherein, if the actual association time does not meet the filtered association time, and / or the actual number of associations does not meet the filtered association time, the propagation path containing the aforementioned target relation edge will be filtered out.

[0157] Further, the attributes corresponding to the aforementioned propagation path include actual association time and actual association count, and the constraints include constraint association time and constraint association count. The process of determining a partial self-loop edge from all self-loop edges of the reached node by comparing the attributes corresponding to the propagation path with the preset constraints includes: comparing the actual association time in the attributes corresponding to the propagation path with the constraint association time to determine a first target behavior group that satisfies the constraint association time; comparing the actual association count in the attributes corresponding to the propagation path with the constraint association count to determine a second target behavior group that satisfies the constraint association count; and determining the intersection of the first target behavior group and the second target behavior group as the partial self-loop edge, wherein the behavior represented by the partial self-loop edge is a target behavior associated with the aforementioned private message.

[0158] Furthermore, determining the risk value corresponding to the target node based on the target behaviors corresponding to the aforementioned partial self-loop edges includes: summing the target behaviors corresponding to the aforementioned partial self-loop edges to obtain the risk value corresponding to the aforementioned target node; or, determining the weights of the target behaviors corresponding to the aforementioned partial self-loop edges, multiplying the target behaviors by their corresponding weights, and then summing the products to obtain the risk value corresponding to the aforementioned target node.

[0159] Furthermore, the processor 1201 is specifically used to: after determining the risk value corresponding to the target node, generate warning information about the target node when it is determined that the risk value corresponding to the target node is greater than a preset threshold.

[0160] Memory 1202 may include one or more computer-readable storage media, which may be non-transitory. Memory 1202 may also include high-speed random access memory and non-volatile memory, such as one or more disk storage devices or flash memory devices. In some embodiments of this specification, the non-transitory computer-readable storage media in memory 1202 is used to store at least one instruction for execution by processor 1201 to implement the methods in the embodiments of this specification.

[0161] In some embodiments, the electronic device 1200 further includes a peripheral device interface 1203 and at least one peripheral device. The processor 1201, memory 1202, and peripheral device interface 1203 can be connected via a bus or signal line. Each peripheral device can be connected to the peripheral device interface 1203 via a bus, signal line, or circuit board. Specifically, the peripheral device includes at least one of a display screen 1204, a camera 1205, and an audio circuit 1206.

[0162] Peripheral interface 1203 can be used to connect at least one input / output (I / O) related peripheral device to processor 1201 and memory 1202. In some embodiments of this specification, processor 1201, memory 1202, and peripheral interface 1203 are integrated on the same chip or circuit board; in some other embodiments of this specification, any one or two of processor 1201, memory 1202, and peripheral interface 1203 can be implemented on separate chips or circuit boards. This specification does not specifically limit the embodiments in this regard.

[0163] Display screen 1204 is used to display a user interface (UI). The UI may include graphics, text, icons, videos, and any combination thereof. When display screen 1204 is a touch display screen, it also has the ability to collect touch signals on or above its surface. These touch signals can be input as control signals to processor 1201 for processing. In this case, display screen 1204 can also be used to provide virtual buttons and / or a virtual keyboard, also known as soft buttons and / or a soft keyboard. In some embodiments of this specification, there may be one display screen 1204, which serves as the front panel of electronic device 1200; in other embodiments, there may be at least two display screens 1204, respectively disposed on different surfaces of electronic device 1200 or in a folded design; in still other embodiments, display screen 1204 may be a flexible display screen, disposed on a curved or folded surface of electronic device 1200. Furthermore, display screen 1204 may also be configured as a non-rectangular irregular shape, i.e., a non-rectangular screen. The display screen 1204 can be made of materials such as liquid crystal display (LCD) and organic light-emitting diode (OLED).

[0164] Camera 1205 is used to capture images or videos. Optionally, camera 1205 includes a front-facing camera and a rear-facing camera. Typically, the front-facing camera is located on the front panel of the electronic device, and the rear-facing camera is located on the back of the electronic device. In some embodiments, there are at least two rear-facing cameras, which are any one of a main camera, a depth-sensing camera, a wide-angle camera, and a telephoto camera, to achieve background blurring by fusion of the main camera and the depth-sensing camera, panoramic shooting by fusion of the main camera and the wide-angle camera, virtual reality (VR) shooting, or other fusion shooting functions. In some embodiments of this specification, camera 1205 may also include a flash. The flash can be a single-color temperature flash or a dual-color temperature flash. A dual-color temperature flash refers to a combination of a warm light flash and a cool light flash, which can be used for light compensation at different color temperatures.

[0165] The audio circuit 1206 may include a microphone and a speaker. The microphone is used to collect sound waves from the user and the environment, and convert the sound waves into electrical signals that are input to the processor 1201 for processing. For stereo sound acquisition or noise reduction purposes, there may be multiple microphones, each located in a different part of the electronic device 1200. The microphone may also be an array microphone or an omnidirectional microphone.

[0166] Power supply 1207 is used to supply power to various components in electronic device 1200. Power supply 1207 can be AC ​​power, DC power, a disposable battery, or a rechargeable battery. When power supply 1207 includes a rechargeable battery, the rechargeable battery can be a wired rechargeable battery or a wireless rechargeable battery. A wired rechargeable battery is a battery that is charged via a wired line, and a wireless rechargeable battery is a battery that is charged via a wireless coil. The rechargeable battery can also be used to support fast charging technology.

[0167] The block diagrams of the electronic device shown in the embodiments of this specification do not constitute a limitation on the electronic device 1200. The electronic device 1200 may include more or fewer components than shown, or combine certain components, or use different component arrangements.

[0168] In the description of this specification, it should be understood that the terms "first," "second," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance. Those skilled in the art can understand the specific meaning of these terms in this specification based on the specific circumstances. Furthermore, in the description of this specification, unless otherwise stated, "multiple" means two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship.

[0169] This specification also provides a computer-readable storage medium storing instructions that, when executed on a computer or processor, cause the computer or processor to perform one or more steps in the above embodiments. If the constituent modules of the above-described availability determination device are implemented as software functional units and sold or used as independent products, they can be stored in the above-described computer-readable storage medium.

[0170] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions. When these computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this specification are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in or transmitted through a computer-readable storage medium. The computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, Digital Subscriber Line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium accessible to a computer or a data storage device such as a server or data center that integrates one or more available media. The aforementioned available media can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., Digital Versatile Discs (DVDs)), or semiconductor media (e.g., Solid State Disks (SSDs)).

[0171] It should be noted that the above description describes specific embodiments of this specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims may be performed in a different order than that shown in the embodiments and still achieve the desired results. Furthermore, the processes depicted in the drawings do not necessarily require the specific or sequential order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0172] The above description is merely a specific embodiment of this specification, but the scope of protection of this specification is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this specification should be included within the scope of protection of this specification. Therefore, equivalent variations made in accordance with the claims of this specification are still within the scope of this specification.

Claims

1. A data processing method, wherein, The method includes: For a target node in a graph relation network, determine the propagation path of the private message stored in the target node, wherein the private message is sent by a message passing mechanism based on the synchronous parallel BSP computing model, and the propagation path includes reaching nodes; For the relational edges included in the propagation path, the association time and / or the number of associations between the two nodes connected by the relational edge are determined as the attribute corresponding to the relational edge; wherein, determining the actual association time and / or the actual number of associations between the two nodes connected by the relational edge as the attribute corresponding to the relational edge includes: when the propagation path sequentially includes the target node, an intermediate node, and the reach node, determining the actual association time and / or the actual number of associations between the target node and the intermediate node as the attribute corresponding to the first relational edge, and determining the actual association time and / or the actual number of associations between the intermediate node and the reach node as the attribute corresponding to the second relational edge; The attributes corresponding to the propagation path are determined based on the attributes corresponding to the relation edges contained in the propagation path; wherein, determining the attributes corresponding to the propagation path based on the attributes corresponding to the relation edges contained in the propagation path includes: determining the attributes corresponding to the propagation path based on the attributes corresponding to the first relation edge and the attributes corresponding to the second relation edge; By comparing the attributes corresponding to the propagation path with the preset constraints, some self-loop edges are identified among all the self-loop edges of the reached node. Based on the target behaviors corresponding to the partial self-loop edges, the risk value corresponding to the target node is determined.

2. The method according to claim 1, wherein, The method further includes: Based on the message passing mechanism of the BSP computing model, private messages corresponding to each node are saved. Based on the relationship data between nodes, the relationship edges between nodes are determined, and the self-loop edges of nodes are determined based on the support data of nodes, thus obtaining the graph relationship network; The nodes include: users and media, wherein the media includes one or more of devices, certificates, and identification numbers.

3. The method according to claim 1, wherein, The method further includes: In the graph relationship network, a subgraph is determined, wherein the subgraph includes: an initiating node corresponding to a risk initiator and an reaching node corresponding to a risk triggerer, wherein the reaching node includes one or more; For a target node in a graph relational network, determining the propagation path of the private messages stored by the target node includes: The starting node in the subgraph is determined as the target node; The path through which the private message is propagated from the target node to the reach node is determined as the propagation path.

4. The method according to claim 1, wherein, The step of determining the association time and / or association count between the two nodes connected by the relationship edge in the propagation path as the attribute corresponding to the relationship edge further includes: When the propagation path sequentially includes the target node and the reach node, the actual association time and / or the actual association count between the target node and the reach node are determined as the attribute corresponding to the relationship edge; The step of determining the attribute corresponding to the propagation path based on the attribute corresponding to the relation edge contained in the propagation path further includes: The attribute corresponding to the relation edge is determined as the attribute corresponding to the propagation path.

5. The method according to claim 1, wherein, Determining the attribute corresponding to the propagation path based on the attribute corresponding to the first relation edge and the attribute corresponding to the second relation edge includes: Based on the actual association time in the attributes corresponding to the first relation edge and the actual association time in the attributes corresponding to the second relation edge, determine the actual association time in the attributes of the propagation path; and / or, The actual number of associations in the attributes of the propagation path is determined based on the actual number of associations in the attributes of the first relation edge and the actual number of associations in the attributes of the second relation edge.

6. The method according to claim 1, wherein, For a target node in a graph relational network, determining the propagation path of the private messages stored by the target node includes: For any relation edge in the subgraph, compare the attribute corresponding to the relation edge with the preset filtering conditions; If the attribute corresponding to the relation edge does not meet the filtering condition, the propagation path containing the relation edge is filtered out. The remaining propagation path after filtering is the propagation path of the private message stored by the target node, and the attribute corresponding to it is compared with the constraint condition.

7. The method according to claim 6, wherein, The attributes corresponding to the relation edge include the actual association time and the actual number of associations, and the filtering conditions include filtering association time and filtering the number of associations; The comparison of the attributes corresponding to the relation edges with the preset filtering conditions includes: For a target relation edge, compare the actual association time and the filtered association time in the attributes corresponding to the target relation edge, and compare the actual association count and the filtered association count in the attributes corresponding to the target relation edge; Where the actual association time does not meet the filtering association time, and / or the actual association count does not meet the filtering association count, the propagation path containing the target relationship edge will be filtered out.

8. The method according to any one of claims 1 to 7, wherein, The attributes corresponding to the propagation path include the actual association time and the actual association count, and the constraints include the constraint association time and the constraint association count. The method involves comparing the attributes corresponding to the propagation path with preset constraints to identify a portion of the self-loop edges among all self-loop edges of the reached node, including: By comparing the actual association time and the constraint association time in the attributes corresponding to the propagation path, a first target behavior group that satisfies the constraint association time is determined; By comparing the actual number of associations in the attributes corresponding to the propagation path with the number of constraint associations, a second target behavior group that satisfies the number of constraint associations is determined; The intersection of the first target behavior group and the second target behavior group is determined as the partial self-loop edge, wherein the behavior represented by the partial self-loop edge is the target behavior associated with the private message.

9. The method according to any one of claims 1 to 7, wherein, The step of determining the risk value corresponding to the target node based on the target behaviors corresponding to the partial self-loop edges includes: Sum the target behaviors corresponding to the partial self-loop edges to obtain the risk value corresponding to the target node; or, Determine the weights of the target behaviors corresponding to the partial self-loop edges, multiply the target behaviors by their corresponding weights, and then sum them to obtain the risk value corresponding to the target node.

10. The method according to any one of claims 1 to 7, wherein, After determining the risk value corresponding to the target node, the method further includes: If the risk value corresponding to the target node is determined to be greater than a preset threshold, a warning message is generated for the target node.

11. A data processing apparatus, wherein, The device includes: The path determination module is used to determine the propagation path of the private message stored in the target node in the graph relation network, wherein the private message is sent by the message passing mechanism based on the synchronous parallel BSP computing model, and the propagation path includes the reaching node. The self-loop edge filtering module is used to identify some self-loop edges among all self-loop edges of the reached node by comparing the attributes corresponding to the propagation path and the preset constraints. The risk value determination module is used to determine the risk value corresponding to the target node based on the target behaviors corresponding to the partial self-loop edges; The device is further configured to: for a relation edge included in the propagation path, determine the association time and / or the number of associations between two nodes connected by the relation edge as the attribute corresponding to the relation edge; wherein, determining the actual association time and / or the actual number of associations between two nodes connected by the relation edge included in the propagation path as the attribute corresponding to the relation edge includes: when the propagation path sequentially includes the target node, an intermediate node, and the reach node, determining the actual association time and / or the actual number of associations between the target node and the intermediate node as the attribute corresponding to the first relation edge, and determining the actual association time and / or the actual number of associations between the intermediate node and the reach node as the attribute corresponding to the second relation edge; The attributes corresponding to the propagation path are determined based on the attributes corresponding to the relation edges contained in the propagation path; wherein, determining the attributes corresponding to the propagation path based on the attributes corresponding to the relation edges contained in the propagation path includes: determining the attributes corresponding to the propagation path based on the attributes corresponding to the first relation edge and the attributes corresponding to the second relation edge.

12. A computer-readable storage medium storing instructions, wherein, When the instructions are executed on a computer or processor, the computer or processor performs the data processing method as described in any one of claims 1 to 10.

13. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein, When the processor executes the computer program, it implements the data processing method as described in any one of claims 1 to 10.

14. A computer program product containing instructions, wherein, When the computer program product is run on a computer or processor, it causes the computer or processor to perform the data processing method as described in any one of claims 1 to 10.

Images