A method and device for screening association relations based on a graph

Abstract

Embodiments of the present application provide a kind of based on atlas's association relationship screening method and device, applied in artificial intelligence technical field, comprising: obtaining historical transaction data, and based on historical transaction data, constructs transaction association graph;Based on the transaction time of each transaction in historical transaction data, transaction association graph is split into multiple association subgraphs, each association subgraph includes a time period node and associated user node and transaction object node;Multiple association subgraphs are clustered, and multiple clustering sets are obtained;For each clustering set, multiple association subgraphs in clustering set are atlas fusion, obtain sequence fusion subgraph, and determine overlapping target association subgraph from sequence fusion subgraph.The behavior sequence of user and the association relationship of atlas are fused in the present application, the common behavior data is clustered by the method of clustering, can excavate strong centricity user transaction behavior, improve the analysis efficiency and accuracy to point arbitrage behavior.

Description

TECHNICAL FIELD

[0001] Embodiments of the present application relate to the technical field of artificial intelligence, and in particular to a graph-based association relationship screening method and device. BACKGROUND

[0002] In the exchange and use scenarios of points and consumption coupons, malicious users often use points and consumption coupons for arbitrage. When judging whether the exchange and use behavior of a user is malicious behavior, the association of multi-link data is involved, and the association relationship of the data is relatively complex.

[0003] Under the related technology, when the association of multi-link data is determined by analyzing the time sequence data of the exchange and use behavior of a user, the accuracy and efficiency are low. SUMMARY

[0004] Embodiments of the present application provide a graph-based association relationship screening method and device for screening point arbitrage behavior in user transaction behavior.

[0005] In a first aspect, the embodiments of the present application provide a graph-based association relationship screening method, comprising:

[0006] Obtaining historical transaction data, and constructing a transaction association graph based on the historical transaction data, the transaction association graph comprising a plurality of user nodes and associated transaction object nodes;

[0007] Based on the transaction time of each transaction in the historical transaction data, the transaction association graph is split into a plurality of association subgraphs, each association subgraph comprising a time period node and associated user nodes and transaction object nodes;

[0008] Clustering the plurality of association subgraphs to obtain a plurality of clustering sets;

[0009] For each clustering set, performing graph fusion on the plurality of association subgraphs in the clustering set to obtain a sequence fusion subgraph, and determining an overlapping target association subgraph from the sequence fusion subgraph.

[0010] The present application fuses the behavior sequence of a user and the association relationship of a graph, splits a complex graph, and clusters behavior data with commonalities by a clustering method to form a clustering combination. While retaining more fine-grained information, the present application can mine strong centralized user transaction behavior, so that it has a better analysis effect in point arbitrage behavior detection and analysis, and improves the efficiency and accuracy of analysis.

[0011] Optionally, for any two association subgraphs, if the number of the same user nodes in the two association subgraphs is greater than or equal to a preset threshold, the two association subgraphs are divided into one clustering set.

[0012] Optionally, the transaction association graph is split into a plurality of association sub-graphs based on transaction time and transaction object type of each transaction in the historical transaction data.

[0013] Optionally, the lines in the plurality of association sub-graphs are encoded in a random walk manner to obtain an embedding matrix, and each row of encoding sequence in the embedding matrix includes a time sequence and a transaction object sequence corresponding to a user node.

[0014] A plurality of combined sequences are obtained by clustering a plurality of rows of encoding sequences in the embedding matrix.

[0015] The plurality of association sub-graphs are divided into a plurality of clustering sets based on the plurality of combined sequences.

[0016] Optionally, time period nodes of the plurality of association sub-graphs in the clustering set are connected to obtain a sequence fusion sub-graph.

[0017] Optionally, at least one transaction sequence is obtained from the sequence fusion sub-graph, and the transaction sequence includes a user node, a transaction object node, and a connection relationship between the transaction node and the transaction object node.

[0018] The target association sub-graph is obtained based on the at least one transaction sequence.

[0019] In a second aspect, an embodiment of the present application provides a graph-based association relationship screening device, including:

[0020] An acquisition module is configured to acquire historical transaction data and construct a transaction association graph based on the historical transaction data, the transaction association graph including a plurality of user nodes and associated transaction object nodes.

[0021] A processing module is configured to split the transaction association graph into a plurality of association sub-graphs based on transaction time of each transaction in the historical transaction data, and each association sub-graph includes a time period node and associated user nodes and transaction object nodes.

[0022] The processing module is further configured to cluster the plurality of association sub-graphs to obtain a plurality of clustering sets.

[0023] The processing module is further configured to, for each clustering set, perform graph fusion on the plurality of association sub-graphs in the clustering set to obtain a sequence fusion sub-graph, and determine an overlapping target association sub-graph from the sequence fusion sub-graph.

[0024] The application fuses the association relationship of the behavior sequence of the user and the graph, splits the complex graph, clusters the common behavior data by the clustering method, and forms a cluster combination. While retaining more fine-grained information, the strong concentrated user transaction behavior can be mined, so that the analysis effect is better in the point arbitrage behavior detection and analysis, and the analysis efficiency and accuracy are improved.

[0025] Optionally, the processing module is specifically configured to:

[0026] For any two associated subgraphs, if the number of same user nodes in the two associated subgraphs is greater than or equal to a preset threshold, the two associated subgraphs are divided into one cluster set.

[0027] Optionally, the processing module is specifically configured to:

[0028] Based on the transaction time and the transaction object type of each transaction in the historical transaction data, the transaction association graph is split into a plurality of associated subgraphs.

[0029] Optionally, the processing module is specifically configured to:

[0030] The lines in the plurality of associated subgraphs are encoded in a random walk manner to obtain an embedding matrix, and each row of the embedding matrix includes a time sequence and a transaction object sequence corresponding to a user node.

[0031] A plurality of combination sequences are obtained by clustering a plurality of encoding sequences in the embedding matrix.

[0032] Based on the plurality of combination sequences, the plurality of associated subgraphs are divided into a plurality of cluster sets.

[0033] Optionally, the processing module is specifically configured to:

[0034] The time period nodes of the plurality of associated subgraphs in the cluster set are connected to obtain the sequence fusion subgraph.

[0035] Optionally, the processing module is specifically configured to:

[0036] At least one overlapping transaction sequence is obtained from the sequence fusion subgraph, and the transaction sequence includes a user node, a transaction object node, and a connection relationship between the transaction node and the transaction object node.

[0037] Based on the at least one transaction sequence, the target associated subgraph is obtained.

[0038] In a third aspect, the embodiments of the present application provide a computer device, comprising a memory, a processor, and a computer program stored in the memory and capable of running on the processor, wherein the processor executes the graph-based association relationship screening method according to any of the first aspect.

[0039] In a fourth aspect, the embodiments of the present application provide a computer readable storage medium, which stores a computer program capable of being executed by a computer device, and when the program runs on the computer device, the computer device executes the graph-based association relationship screening method according to any of the first aspect.

[0040] The present application fuses the behavior sequence of a user and the association relationship of a graph, splits a complex graph, and clusters the behavior data with commonality by a clustering method to form a clustering combination. While retaining more fine-grained information, the present application can mine strong concentrated user transaction behaviors, so that the analysis effect is better in point arbitrage behavior detection and analysis, and the analysis efficiency and accuracy are improved. BRIEF DESCRIPTION OF DRAWINGS

[0041] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0042] Figure 1 A structural schematic diagram of a system architecture diagram is provided for the embodiments of the present application;

[0043] Figure 2 A flowchart of a graph-based association relationship screening method is provided for the embodiments of the present application;

[0044] Figure 3 A schematic diagram of a transaction association graph is provided for the embodiments of the present application;

[0045] Figure 4 A schematic diagram of an association subgraph is provided for the embodiments of the present application;

[0046] Figure 5 A schematic diagram of the association between a time node and a user node is provided for the embodiments of the present application;

[0047] Figure 6 A schematic diagram of a clustering set is provided for the embodiments of the present application;

[0048] Figure 7 A schematic diagram of a transaction association graph is provided for the embodiments of the present application;

[0049] Figure 8 A schematic diagram of an associated subgraph provided for an embodiment of the present application;

[0050] Figure 9 A schematic diagram of an associated subgraph provided for an embodiment of the present application;

[0051] Figure 10 A schematic diagram of a sequence fusion subgraph provided for an embodiment of the present application;

[0052] Figure 11 A schematic diagram of a sequence fusion subgraph provided for an embodiment of the present application;

[0053] Figure 12 A schematic diagram of a target associated subgraph provided for an embodiment of the present application;

[0054] Figure 13 A schematic diagram of a target associated subgraph provided for an embodiment of the present application;

[0055] Figure 14 A schematic diagram of a structure of an associated relationship screening device based on a graph provided for an embodiment of the present application;

[0056] Figure 15 A schematic diagram of a structure of a computing device provided for an embodiment of the present application. DETAILED DESCRIPTION

[0057] In order to make the objectives, technical solutions and beneficial effects of the present application clearer, the present application is further described in detail below in combination with the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application, and are not used to limit the present application.

[0058] Referring to Figure 1 is a system architecture diagram used by an embodiment of the present application, the system architecture 100 includes a terminal device 101 and a server 102, wherein the terminal device 101 is used to collect behaviors of users exchanging points and using consumption coupons; the terminal device 101 can be a smart phone, a tablet computer, a notebook computer, a desktop computer, etc., but is not limited thereto.

[0059] The server 102 receives the user exchange points and the behavior of using the consumption coupons collected by the terminal device 101, and filters the consumption transaction group cluster with strong concentration based on the user exchange points and the behavior of using the consumption coupons in combination with the association relationship of the graph. The server 102 can be a physical server, a server cluster composed of multiple physical servers or a distributed system, or a cloud server providing cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDN), and basic cloud computing services such as big data and artificial intelligence platforms.

[0060] The terminal device 101 and the server 102 can be directly or indirectly connected through wired or wireless communication, which is not limited in the present application.

[0061] Based on the above system architecture, Figure 2 An exemplary method for filtering based on the association relationship of the graph is shown, including the following steps:

[0062] In step S201, historical transaction data is obtained, and a transaction association graph is constructed based on the historical transaction data. The transaction association graph includes a plurality of user nodes and associated transaction object nodes.

[0063] Specifically, for the historical transactions between the plurality of user nodes and the transaction object nodes, a transaction association graph is constructed. The transaction object node can be a merchant node that transacts with the user, or a coupon node used by the user in the transaction. The connection between a user node and a transaction object node represents a historical transaction, and a historical transaction is represented as a relationship edge in the transaction association graph. The association relationship edge also represents the attribute information of the transaction between the user and the transaction object, and the attribute information can be the total transaction amount, the total transaction number, and the high-frequency transaction segment information.

[0064] For example, in the obtained historical transaction data, user 1 transacts with merchant 1 and merchant 2 respectively; user 2 transacts with merchant 1 and merchant 2 respectively; and user 3 transacts with merchant 1 and merchant 2 respectively. According to the above historical transaction data, a transaction association graph is constructed as shown in FIG. 1, taking user 1, user 2, user 3, merchant 1, and merchant 2 as nodes. Figure 3

[0065] In step S202, based on the transaction time of each transaction in the historical transaction data, the transaction association graph is split into a plurality of association subgraphs, and each association subgraph includes a time period node and associated user nodes and transaction object nodes.

[0066] ​Specifically, based on the time periods of historical transactions, the transaction association graph is split into multiple association subgraphs, each including a time period node. An association subgraph represents a transaction between a user node and a transaction object node within that time period.

[0067] For example, as Figure 3 The transaction relationship diagram shown is broken down according to the time periods of transactions between User 1, User 2, User 3, Merchant 1, and Merchant 2, resulting in the following: Figure 4 The diagram shows three related subgraphs. Taking the first subgraph as an example, it represents that within the first time period (1H), user 1 transacted with merchant 2, user 2 transacted with merchant 1, and user 3 transacted with merchant 1. The second and third subgraphs represent the same content.

[0068] Step S203: Cluster the multiple associated subgraphs to obtain multiple cluster sets.

[0069] Step S204: For each cluster set, perform graph fusion on multiple association subgraphs in the cluster set to obtain a sequence fused subgraph, and determine the overlapping target association subgraphs from the sequence fused subgraph.

[0070] Specifically, for each cluster set, multiple association subgraphs within the cluster set are fused using graph fusion to obtain a sequence fused subgraph, which is obtained by connecting multiple association subgraphs. Based on the obtained sequence fused subgraph, overlapping target association subgraphs are determined.

[0071] This application integrates the correlation between user behavior sequences and graphs, decomposes complex graphs, and clusters common behavioral data to form cluster combinations. While retaining finer-grained information, it can uncover highly concentrated user transaction behaviors, resulting in superior analytical performance in the detection and analysis of point arbitrage behavior, thus improving the efficiency and accuracy of the analysis.

[0072] In some embodiments, multiple associated subgraphs are clustered to obtain multiple cluster sets, including: for any two associated subgraphs, if the number of the same user nodes in the two associated subgraphs is greater than or equal to a preset threshold, then the two associated subgraphs are divided into a cluster set.

[0073] Specifically, the preset threshold represents the number of identical user nodes, and the number of identical nodes is set according to the actual application scenario. If two related subgraphs contain the same user nodes, they are considered to have commonalities. If the number of identical user nodes in two related subgraphs is greater than or equal to the preset threshold, the two related subgraphs are clustered, that is, they are assigned to a single cluster set.

[0074] For example, the three association subgraphs as shown in Figure 4 are divided according to the time node and the user node information to obtain the three association graphs of the time node and the user node as shown in Figure 5 . The preset threshold of the same user of the association subgraph is set to 3. In the first time period, users 1, 2 and 3 have transactions. In the second time period, users 1 and 2 have transactions. In the third time period, users 1, 2 and 3 have transactions. In the first time period and the second time period, the number of the same user nodes is equal to 3. Therefore, the association subgraphs of the first time period and the second time period are clustered to obtain a cluster set, and the cluster set is as shown in Figure 6 .

[0075] In some embodiments, the transaction association graph is split into multiple association subgraphs based on the transaction time of each transaction in the historical transaction data, including: splitting the transaction association graph into multiple association subgraphs based on the transaction time and the transaction object type of each transaction in the historical transaction data.

[0076] Specifically, the transaction object node includes a type attribute, and the transaction association graph is split into multiple association subgraphs according to the transaction time and the transaction object type.

[0077] For example, if the transaction object node is a coupon, the transaction object type is the category attribute of the coupon, that is, the transaction object type is divided into audio and payment categories, and the transaction association graph as shown in Figure 7 is composed of 5 user nodes, 3 coupon nodes and 2 type attribute nodes. User 1 has a coupon 1 and a coupon 2 redemption transaction. User 2 has a coupon 1 and a coupon 2 redemption transaction. User 3 has a coupon 1, a coupon 2 and a coupon 3 redemption transaction. User 4 has a coupon 2 and a coupon 3 redemption transaction. User 5 has a coupon 3 redemption transaction.

[0078] The transaction association graph as shown in Figure 7 is split according to the transaction time and the transaction object type to obtain multiple association subgraphs as shown in Figure 8 .

[0079] In some embodiments, the multiple association subgraphs are clustered to obtain multiple cluster sets, including: encoding the lines in the multiple association subgraphs in a random walk manner to obtain an embedding matrix, each row of the embedding matrix including: a time sequence and a transaction object sequence corresponding to a user node; clustering the multiple rows of the embedding matrix to obtain multiple combination sequences; and dividing the multiple association subgraphs into multiple cluster sets based on the multiple combination sequences.

[0080] Specifically, the random walk method refers to starting from a point in the associated subgraph, randomly selecting a neighboring node, moving to that neighbor, and then repeating this process from the current node, forming a node sequence, also known as a node path. One-hot encoding is then used to encode each path in the associated subgraph, obtaining an embedding matrix. Each row of the embedding matrix is ​​encoded by a time series (X... time The sequence consists of a coupon and a transaction object sequence. If the transaction object is a coupon, then the transaction object sequence can represent the coupon category (X). category The time series is divided into 12 time periods with 2-hour intervals, and the time series is encoded using 12-dimensional One-Hot encoding. The transaction object series is divided into 7 categories, and the transaction object series is encoded using 7-dimensional One-Hot encoding. The 12-dimensional time series and the 7-dimensional transaction object series are concatenated to obtain a 19-dimensional sequence. This sequence is denoted as X, and X equals X... time concat X category This encoding method allows for the simultaneous representation of time and category information within a single sequence. Clustering the multi-row encoded sequences in the embedding matrix yields multiple combined sequences, each denoted as X, where X is a combination of multiple row encoded sequences. i seq =X i concat X j , where X i X j This is the encoded sequence in a combined sequence. Then, based on multiple combined sequences, the multiple association subgraphs are divided into multiple cluster sets.

[0081] For example, using a random walk method to... Figure 8 The lines in the multiple associated subgraphs shown are encoded to obtain the following: Figure 9 The embedding matrix shown contains multiple rows of encoded sequences. Each row contains a 12-dimensional time series and a 7-dimensional transaction object sequence, where the transaction object is the category of coupons. In this embodiment, User 1 and User 3 have the same time series and the same category sequence. Therefore, the encoded sequences corresponding to User 1 and User 3 are clustered to obtain a combined sequence. Based on this combined sequence, the corresponding multiple association subgraphs are divided into multiple cluster sets.

[0082] In some embodiments, graph fusion is performed on multiple associated subgraphs in a cluster set to obtain a sequence fusion subgraph, including: establishing connections between time period nodes of multiple associated subgraphs in a cluster set to obtain a sequence fusion subgraph.

[0083] Specifically, the time period nodes of the multiple associated subgraphs included in each cluster set are connected to obtain a sequence fusion subgraph. The sequence fusion subgraph includes more refined exchange information. For example, as shown in the cluster set in Figure 6 , the cluster set includes two associated subgraphs, the time period nodes of the two associated subgraphs are connected, that is, the first time period node and the third time period node are connected, to obtain the sequence fusion subgraph as shown in Figure 10 . For another example, as shown in the cluster set in Figure 9 , the cluster set includes two associated subgraphs, the time period nodes of the two associated subgraphs are connected, that is, the first time period node and the second time period node are connected, to obtain the sequence fusion subgraph as shown in Figure 11 .

[0084] In some embodiments, determining an overlapping target associated subgraph from the sequence fusion subgraph includes: obtaining at least one transaction sequence that overlaps from the sequence fusion subgraph, the transaction sequence including a user node, a transaction object node, and a connection relationship between the transaction node and the transaction object node; and obtaining the target associated subgraph based on the at least one transaction sequence.

[0085] Specifically, the overlapping transaction sequence is determined from the sequence fusion subgraph, the transaction sequence including a user node, a transaction object node, and a connection relationship between the transaction node and the transaction object node, and the target associated subgraph is obtained based on the obtained transaction sequence. The overlapping transaction sequence can finely divide the interaction cluster of the user node and the transaction object node.

[0086] For example, the overlapping part in the sequence fusion subgraph as shown in Figure 10 is extracted to obtain the target associated subgraph as shown in Figure 12 . If the overlapping part in the sequence fusion subgraph as shown in Figure 11 is extracted to obtain the target associated subgraph as shown in Figure 13 .

[0087] Based on the same technical concept, the embodiments of the present application provide a structural schematic diagram of a graph-based association relationship screening device, as shown in Figure 14 , the device 1400 includes:

[0088] The obtaining module 1401 is configured to obtain historical transaction data, and construct a transaction association graph based on the historical transaction data, the transaction association graph including multiple user nodes and associated transaction object nodes.

[0089] The processing module 1402 is configured to split the transaction association graph into multiple associated subgraphs based on the transaction time of each transaction in the historical transaction data, each associated subgraph including a time period node and associated user nodes and transaction object nodes.

[0090] The processing module 1402 is further configured to cluster the plurality of association subgraphs to obtain a plurality of cluster sets.

[0091] The processing module 1402 is further configured to, for each cluster set, perform graph fusion on the plurality of association subgraphs in the cluster set to obtain a sequence fusion subgraph, and determine an overlapping target association subgraph from the sequence fusion subgraph.

[0092] The application fuses the behavior sequence of the user and the association relationship of the graph, splits the complex graph, clusters the common behavior data by the clustering method, and forms a cluster combination. While retaining more fine-grained information, the application can mine strong concentrated user transaction behaviors, so that the analysis effect is better in the point arbitrage behavior detection and analysis, and the analysis efficiency and accuracy are improved.

[0093] Optionally, the processing module 1402 is specifically configured to:

[0094] For any two association subgraphs, if the number of the same user nodes in the two association subgraphs is greater than or equal to a preset threshold, the two association subgraphs are divided into one cluster set.

[0095] Optionally, the processing module 1402 is specifically configured to:

[0096] The transaction association graph is split into a plurality of association subgraphs based on the transaction time and the transaction object type of each transaction in the historical transaction data.

[0097] Optionally, the processing module 1402 is specifically configured to:

[0098] The lines in the plurality of association subgraphs are encoded in a random walk manner to obtain an embedding matrix, and each row of the embedding matrix includes a time sequence and a transaction object sequence corresponding to a user node.

[0099] A plurality of combination sequences are obtained by clustering a plurality of encoding sequences in the embedding matrix.

[0100] The plurality of association subgraphs are divided into a plurality of cluster sets based on the plurality of combination sequences.

[0101] Optionally, the processing module 1402 is specifically configured to:

[0102] The time period nodes of the plurality of association subgraphs in the cluster set are connected to obtain the sequence fusion subgraph.

[0103] Optionally, the processing module 1402 is specifically configured to:

[0104] Obtain at least one overlapping transaction sequence from the sequence fusion subgraph, wherein the transaction sequence includes a user node, a transaction object node, and the connection relationship between the transaction node and the transaction object node;

[0105] Based on the at least one transaction sequence, the target correlation subgraph is obtained.

[0106] This application integrates the correlation between user behavior sequences and graphs, decomposes complex graphs, and clusters common behavioral data to form cluster combinations. While retaining finer-grained information, it can uncover highly concentrated user transaction behaviors, resulting in superior analytical performance in the detection and analysis of point arbitrage behavior, thus improving the efficiency and accuracy of the analysis.

[0107] Based on the same technical concept, embodiments of this application provide a computer device, such as... Figure 15 As shown, it includes at least one processor 1501 and a memory 1502 connected to at least one processor. In this embodiment, the specific connection medium between the processor 1501 and the memory 1502 is not limited. Figure 15 Taking the connection between processor 1501 and memory 1502 via a bus as an example, the bus can be divided into address bus, data bus, control bus, etc.

[0108] In this embodiment of the application, the memory 1502 stores instructions that can be executed by at least one processor 1501. By executing the instructions stored in the memory 1502, at least one processor 1501 can perform the steps of the above-described graph-based association filtering method.

[0109] The processor 1501 serves as the control center of the computer device, connecting various parts of the device via interfaces and lines. It filters out highly centralized consumer transaction groups by running or executing instructions stored in the memory 1502 and accessing data stored in the memory 1502. Optionally, the processor 1501 may include one or more processing units. The processor 1501 may integrate an application processor and a modem processor. The application processor primarily handles the operating system, user interface, and applications, while the modem processor primarily handles wireless communication. It is understood that the modem processor may not be integrated into the processor 1501. In some embodiments, the processor 1501 and the memory 1502 may be implemented on the same chip; in other embodiments, they may be implemented on separate chips.

[0110] The processor 1501 can be a general processor, such as a central processing unit (CPU), a digital signal processor, an application specific integrated circuit (ASIC), a field programmable gate array or other programmable logic device, a discrete gate or transistor logic component, a discrete hardware component, and can implement or execute the methods, steps and logic block diagrams disclosed in the embodiments of the present application. The general processor can be a microprocessor or any conventional processor. The steps of the method disclosed in combination with the embodiments of the present application can be directly embodied as hardware processor execution or executed by a combination of hardware and software modules in the processor.

[0111] The memory 1502 is a non-volatile computer readable storage medium, which can be used to store non-volatile software programs, non-volatile computer executable programs and modules. The memory 1502 can include at least one type of storage medium, such as flash memory, hard disk, multimedia card, card type memory, random access memory (RAM), static random access memory (SRAM), programmable read only memory (PROM), read only memory (ROM), electrically erasable programmable read only memory (EEPROM), magnetic storage, magnetic disk, optical disk, etc. The memory 1502 is any other medium capable of carrying or storing desired program code in the form of instructions or data structures and capable of being accessed by a computer, but is not limited thereto. The memory 1502 in the embodiments of the present application can also be a circuit or any other device capable of realizing a storage function, used to store program instructions and / or data.

[0112] Based on the same inventive concept, the embodiments of the present application provide a computer readable storage medium storing a computer program executable by a computer device, which, when the program is running on the computer device, causes the computer device to execute the steps of the above-mentioned graph-based correlation relationship screening method.

[0113] Those skilled in the art will appreciate that embodiments of the present application can be devised for a variety of applications. FIG. 1 illustrates an example of a system 100 that can employ an embodiment of the present application. As shown in FIG. 1, system 100 can include a host computer 110 that is configured to communicate via one or more wired or wireless communication links 120 with one or more client devices 130. Host computer 110 can include a processor 112, a storage 114, and a communications interface 116. Processor 112 can include one or more processors, such as one or more general purpose processors (e.g., as described below in connection with FIG. 2). Storage 114 can include one or more non-transitory computer-readable storage medium, such as one or more hard disk drives, flash memories, or the like. Storage 114 can store instructions 118 that are executable by processor 112 to implement a method, such as the method described below in connection with FIG. 3. Communications interface 116 can include one or more communications interfaces, such as an Ethernet interface, a Bluetooth interface, a Wi-Fi interface, or the like. Communications interface 116 can be configured to communicate with one or more client devices 130 via one or more wired or wireless communication links 120, such as one or more Ethernet links, one or more Bluetooth links, one or more Wi-Fi links, or the like.

[0114] The present application is described below in reference to flowcharts and / or block diagrams that illustrate the method, apparatus (system), and computer program product according to the present application. It will be understood that each block of the flowchart and / or block diagrams, and combinations of blocks in the flowchart and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general purpose computer, special purpose computer, embedded processing element or other programmable data processing device to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing device, create means for implementing the functions specified in the flowchart and / or block diagram block or blocks. Figure 1 one or more functions specified in the flowchart and / or block diagram block or blocks. Figure 1 one or more functions specified in the flowchart and / or block diagram block or blocks.

[0115] These computer program instructions can also be stored in a computer- readable memory that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instructions which implement the function specified in the flowchart and / or block diagram block or blocks. Figure 1 one or more functions specified in the flowchart and / or block diagram block or blocks. Figure 1 one or more functions specified in the flowchart and / or block diagram block or blocks.

[0116] These computer program instructions can also be loaded onto a computer or other programmable data processing device to cause a series of operational steps to be performed on the computer or other programmable device to produce a computer-implemented process such that the instructions which execute on the computer or other programmable device provide steps for implementing the functions specified in the flowchart and / or block diagram block or blocks. Figure 1 one or more functions specified in the flowchart and / or block diagram block or blocks. Figure 1 one or more functions specified in the flowchart and / or block diagram block or blocks.

[0117] Obviously, numerous modifications and variations of the present application are possible in light of the above teachings. It is therefore to be understood that within the scope of the appended claims and their equivalents, the application can be practiced otherwise than as specifically described.

Claims

1. A graph-based association filtering method, characterized in that, include: Acquire historical transaction data and construct a transaction association graph based on the historical transaction data. The transaction association graph includes multiple user nodes and associated transaction object nodes. Based on the transaction time of each transaction in the historical transaction data, the transaction association graph is divided into multiple association subgraphs. Each association subgraph includes a time period node and associated user nodes and transaction object nodes. Cluster the multiple association subgraphs to obtain multiple cluster sets, each cluster set containing multiple association subgraphs; For each cluster set, multiple association subgraphs in the cluster set are fused to obtain a sequence fusion subgraph, and at least one overlapping transaction sequence is obtained from the sequence fusion subgraph. The transaction sequence includes user nodes, transaction object nodes and the connection relationship between the user nodes and the transaction object nodes. The transaction sequence is used to divide the interaction clusters of user nodes and transaction object nodes. Based on the at least one transaction sequence, a target correlation subgraph is obtained.

2. The method as described in claim 1, characterized in that, The clustering of the multiple associated subgraphs to obtain multiple cluster sets includes: For any two associated subgraphs, if the number of the same user nodes in the two associated subgraphs is greater than or equal to a preset threshold, then the two associated subgraphs are divided into a cluster set.

3. The method as described in claim 1, characterized in that, Based on the transaction time of each transaction in the historical transaction data, the transaction association graph is divided into multiple association sub-graphs, including: Based on the transaction time and transaction object type of each transaction in the historical transaction data, the transaction association graph is split into multiple association subgraphs.

4. The method as described in claim 3, characterized in that, The clustering of the multiple associated subgraphs to obtain multiple cluster sets includes: The routes in the multiple associated subgraphs are encoded using a random walk method to obtain an embedding matrix. Each row of the encoding sequence in the embedding matrix includes: a time series and a transaction object sequence corresponding to a user node. Clustering the multi-row encoded sequences in the embedding matrix yields multiple combined sequences; Based on the multiple combined sequences, the multiple association subgraphs are divided into multiple cluster sets.

5. The method as described in claim 1, characterized in that, The step of fusing multiple association subgraphs in the cluster set to obtain a sequence fused subgraph includes: The time period nodes of multiple associated subgraphs in the cluster set are connected to obtain the sequence fusion subgraph.

6. A graph-based association filtering device, characterized in that, include: The acquisition module is used to acquire historical transaction data and construct a transaction association graph based on the historical transaction data. The transaction association graph includes multiple user nodes and associated transaction object nodes. The processing module is used to split the transaction association graph into multiple association subgraphs based on the transaction time of each transaction in the historical transaction data. Each association subgraph includes a time period node and associated user nodes and transaction object nodes. The processing module is further configured to cluster the plurality of associated subgraphs to obtain a plurality of cluster sets, each cluster set containing a plurality of associated subgraphs; The processing module is further configured to perform graph fusion on multiple association subgraphs in each cluster set to obtain a sequence fusion subgraph, and obtain at least one overlapping transaction sequence from the sequence fusion subgraph. The transaction sequence includes user nodes, transaction object nodes and the connection relationship between the user nodes and the transaction object nodes. The transaction sequence is used to divide the interaction clusters of user nodes and transaction object nodes. Based on the at least one transaction sequence, a target correlation subgraph is obtained.

7. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps of the method according to any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that, It stores a computer program executable by a computer device, which, when run on the computer device, causes the computer device to perform the steps of the method according to any one of claims 1 to 5.

Images