CN117436882A - Abnormal transaction identification method, device, computer equipment and storage medium - Google Patents

Abnormal transaction identification method, device, computer equipment and storage medium Download PDF

Info

Publication number
CN117436882A
CN117436882A CN202310883373.7A CN202310883373A CN117436882A CN 117436882 A CN117436882 A CN 117436882A CN 202310883373 A CN202310883373 A CN 202310883373A CN 117436882 A CN117436882 A CN 117436882A
Authority
CN
China
Prior art keywords
transaction
account
vertex
abnormal
grade
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202310883373.7A
Other languages
Chinese (zh)
Inventor
梁杰
宋瑞
高童
郭运雷
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Industrial and Commercial Bank of China Ltd ICBC
Original Assignee
Industrial and Commercial Bank of China Ltd ICBC
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Industrial and Commercial Bank of China Ltd ICBC filed Critical Industrial and Commercial Bank of China Ltd ICBC
Priority to CN202310883373.7A priority Critical patent/CN117436882A/en
Publication of CN117436882A publication Critical patent/CN117436882A/en
Pending legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q20/00—Payment architectures, schemes or protocols
    • G06Q20/38—Payment protocols; Details thereof
    • G06Q20/40—Authorisation, e.g. identification of payer or payee, verification of customer or shop credentials; Review and approval of payers, e.g. check credit lines or negative lists
    • G06Q20/401—Transaction verification
    • G06Q20/4016—Transaction verification involving fraud or risk level assessment in transaction processing
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
    • G06Q40/02—Banking, e.g. interest calculation or account maintenance
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
    • Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management

Abstract

The present application relates to an abnormal transaction identification method, apparatus, computer device, storage medium and computer program product ， Relates to the technical field of artificial intelligence. The method comprises the following steps: acquiring a banking transaction data set and constructing a transaction diagram according to the banking transaction data set; the transaction graph comprises a plurality of account vertexes, and the weight of the transaction edge between every two account vertexes is obtained by assigning a value based on the matching degree of the transaction characteristic between the corresponding two accounts and the preset abnormal characteristic; performing iterative computation on the grade scores corresponding to the peaks of each account of the transaction graph until the grade scores are converged; the grade score corresponding to each account vertex is used for measuring the degree of abnormality of the account vertex relative to other account vertices in the transaction diagram; and identifying abnormal transaction results from the plurality of account transactions corresponding to the plurality of account vertices based on the grade scores corresponding to the account vertices during convergence. The method can improve abnormal transaction identificationAccuracy of (3).

Description

Abnormal transaction identification method, device, computer equipment and storage medium
Technical Field
The present application relates to the field of artificial intelligence, and in particular, to a graph algorithm-based anti-fraud method, apparatus, computer device, storage medium, and computer program product.
Background
With the development and popularization of the Internet, the financial payment market is affected, banking business is changed particularly by financial institutions, and the traditional banking counter business is gradually replaced by electronic transaction. However, threats such as information leakage, theft of funds, fraud, etc. are increasing around electronic transaction channels. Conventional anti-fraud systems typically rely on rule-based methods with low detection accuracy for complex and evolving fraud patterns.
Disclosure of Invention
Based on this, it is necessary to provide an abnormal transaction identification method, apparatus, computer device, computer readable storage medium and computer program product for the above technical problem of low detection accuracy of fraudulent patterns.
In a first aspect, the present application provides a method of abnormal transaction identification. The method comprises the following steps:
acquiring a banking transaction data set, and constructing a transaction diagram according to the banking transaction data set; the transaction graph comprises a plurality of account vertexes, and the weight of the transaction edge between every two account vertexes is obtained by assigning a value based on the matching degree of the transaction characteristic between the corresponding two accounts and the preset abnormal characteristic;
performing iterative computation on the grade scores corresponding to the peaks of all accounts of the transaction graph until the grade scores are converged; the grade score corresponding to each account vertex is used for measuring the degree of abnormality of the account vertex relative to other account vertices in the transaction diagram;
And identifying abnormal transaction results from the plurality of account transactions corresponding to the plurality of account vertices based on the grade scores corresponding to each account vertex when the grade scores are converged.
In one embodiment, the iteratively calculating the rank score corresponding to each account vertex of the transaction map until the rank score converges further includes:
determining total number N of account vertices in the transaction graph;
initializing the grade score of each account vertex to be 1/N.
In one embodiment, after the acquiring the banking transaction data set and constructing the transaction map according to the banking transaction data set, the method further includes:
dividing a plurality of account vertices included in the transaction graph to obtain a plurality of disjoint account subsets;
constructing transaction subgraphs corresponding to each account subset; the transaction sub-graph includes account vertices for accounts in the subset of accounts.
In one embodiment, the iteratively calculating the rank score corresponding to each account vertex of the transaction map until the rank score converges includes:
and carrying out iterative computation on a plurality of transaction subgraphs in parallel according to a preset PageRank algorithm until the grade scores corresponding to the account vertexes of the transaction subgraphs are converged.
In one embodiment, the iterative computation of the transaction subgraphs in parallel according to the preset PageRank algorithm until the rank scores corresponding to the account vertices of the transaction subgraphs converge includes:
in the iterative calculation process of the grade score of any account vertex in each transaction sub-graph, obtaining the weight of each transaction edge corresponding to any account vertex, and adding the weights of all transaction edges corresponding to any account vertex to obtain the sum of the weights of the transaction edges corresponding to any account vertex;
and calculating the grade score corresponding to any account vertex based on a PageRank algorithm preset by the weight of each transaction edge of any account vertex and the sum of the weights of the transaction edges, so as to obtain the grade score updated by any account vertex.
In one embodiment, the calculating the rank score corresponding to the any account vertex based on a PageRank algorithm preset by the weight of each transaction edge of the any account vertex and the sum of the weights of the transaction edges to obtain the rank score updated by the any account vertex includes:
obtaining the current grade score of any account vertex;
And updating and calculating the grade score corresponding to any account vertex based on the current grade score of any account vertex, the preset damping coefficient, the weight of each transaction edge of any account vertex and the sum of the weights of the transaction edges, so as to obtain the updated grade score.
In one embodiment, the obtaining the weight of each transaction edge corresponding to the vertex of any account includes:
acquiring the current time as a first time, and acquiring the transaction occurrence time corresponding to each transaction edge of any account vertex as a second time;
obtaining a time attenuation factor corresponding to each transaction edge of any account vertex based on the first time and the second time;
and multiplying the time attenuation factor by the weight of the corresponding transaction edge in the last iteration calculation to obtain the weight of the adjusted transaction edge, wherein the weight of the transaction edge in the iteration calculation is used as the weight of the transaction edge.
In one embodiment, the transaction characteristics between the two accounts include transaction edges including transaction amount, transaction frequency, and transaction flow direction; the preset abnormal characteristics comprise abnormal amount characteristics, abnormal frequency characteristics and abnormal flow direction characteristics; the method further comprises the steps of:
Obtaining a first matching degree of the transaction amount between the two corresponding accounts and the abnormal amount characteristic,
A second matching degree corresponding to the transaction frequency between the two accounts and the abnormal transaction frequency, and a third matching degree corresponding to the transaction flow direction and the abnormal flow direction characteristics between the two accounts;
and determining the weight of the transaction edge between the two corresponding accounts based on the first matching degree, the second matching degree and the third matching degree.
In a second aspect, the present application also provides an abnormal transaction identification apparatus. The device comprises:
the diagram construction module is used for acquiring a bank transaction data set and constructing a transaction diagram according to the bank transaction data set; the transaction graph comprises a plurality of account vertexes, and the weight of the transaction edge between every two account vertexes is obtained by assigning a value based on the matching degree of the transaction characteristic between the corresponding two accounts and the preset abnormal characteristic;
the score calculation module is used for carrying out iterative calculation on the grade scores corresponding to the vertexes of each account of the transaction graph according to a preset PageRank algorithm until the grade scores are converged; the grade score corresponding to each account vertex is used for measuring the degree of abnormality of the account vertex relative to other account vertices in the transaction diagram;
And the abnormality judgment module is used for identifying abnormal transaction results from the plurality of account transactions corresponding to the plurality of account vertices based on the grade scores corresponding to the account vertices when the grade scores are converged.
In a third aspect, the present application also provides a computer device. The computer device comprises a memory storing a computer program and a processor which when executing the computer program performs the steps of:
acquiring a banking transaction data set, and constructing a transaction diagram according to the banking transaction data set; the transaction graph comprises a plurality of account vertexes, and the weight of the transaction edge between every two account vertexes is obtained by assigning a value based on the matching degree of the transaction characteristic between the corresponding two accounts and the preset abnormal characteristic;
performing iterative computation on the grade scores corresponding to the peaks of all accounts of the transaction graph until the grade scores are converged; the grade score corresponding to each account vertex is used for measuring the degree of abnormality of the account vertex relative to other account vertices in the transaction diagram;
and identifying abnormal transaction results from the plurality of account transactions corresponding to the plurality of account vertices based on the grade scores corresponding to each account vertex when the grade scores are converged.
In a fourth aspect, the present application also provides a computer-readable storage medium. The computer readable storage medium having stored thereon a computer program which when executed by a processor performs the steps of:
acquiring a banking transaction data set, and constructing a transaction diagram according to the banking transaction data set; the transaction graph comprises a plurality of account vertexes, and the weight of the transaction edge between every two account vertexes is obtained by assigning a value based on the matching degree of the transaction characteristic between the corresponding two accounts and the preset abnormal characteristic;
performing iterative computation on the grade scores corresponding to the peaks of all accounts of the transaction graph until the grade scores are converged; the grade score corresponding to each account vertex is used for measuring the degree of abnormality of the account vertex relative to other account vertices in the transaction diagram;
and identifying abnormal transaction results from the plurality of account transactions corresponding to the plurality of account vertices based on the grade scores corresponding to each account vertex when the grade scores are converged.
In a fifth aspect, the present application also provides a computer program product. The computer program product comprises a computer program which, when executed by a processor, implements the steps of:
Acquiring a banking transaction data set, and constructing a transaction diagram according to the banking transaction data set; the transaction graph comprises a plurality of account vertexes, and the weight of the transaction edge between every two account vertexes is obtained by assigning a value based on the matching degree of the transaction characteristic between the corresponding two accounts and the preset abnormal characteristic;
performing iterative computation on the grade scores corresponding to the peaks of all accounts of the transaction graph until the grade scores are converged; the grade score corresponding to each account vertex is used for measuring the degree of abnormality of the account vertex relative to other account vertices in the transaction diagram;
and identifying abnormal transaction results from the plurality of account transactions corresponding to the plurality of account vertices based on the grade scores corresponding to each account vertex when the grade scores are converged.
According to the abnormal transaction identification method, the abnormal transaction identification device, the computer equipment, the storage medium and the computer program product, different weights are given according to the matching degree of the transaction characteristics of the transaction edges and the preset abnormal characteristics by constructing the transaction graph comprising a plurality of account vertexes and the transaction edges between the two account vertexes, and the transaction edges with higher weights are more easily screened out when the grade scores corresponding to the account vertexes of the transaction graph are subjected to iterative calculation by the different weights. Iterative calculation is carried out until the grade scores are converged, and based on the grade scores corresponding to the vertexes of each account when the grade scores are converged, abnormal transaction results are obtained through screening by identifying transaction edges corresponding to transaction points of the vertexes of the accounts with higher grade scores, so that the accuracy of abnormal transaction identification in bank transactions is improved.
Drawings
FIG. 1 is a diagram of an application environment for an abnormal transaction identification method in one embodiment;
FIG. 2 is a flow chart of a method for identifying abnormal transactions in one embodiment;
FIG. 3 is a flow chart of a medium fraction calculation step according to one embodiment;
FIG. 4 is a flow chart of a method for calculating a time attenuation factor according to an embodiment;
FIG. 5 is a flowchart of another embodiment of a method for fraud transaction detection based on an optimization graph algorithm;
FIG. 6 is a block diagram of an abnormal transaction identification device in one embodiment;
fig. 7 is an internal structural diagram of a computer device in one embodiment.
Detailed Description
In order to make the objects, technical solutions and advantages of the present application more apparent, the present application will be further described in detail with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the present application.
The abnormal transaction identification method provided by the embodiment of the application can be applied to an application environment shown in fig. 1. Wherein the terminal 102 communicates with the server 104 via a network. The server 104 acquires banking data generated from the terminal 102, and the data storage system may store banking data that the server 104 needs to process. The data storage system may be integrated on the server 104 or may be located on a cloud or other network server. The server 104 constructs a transaction graph according to the bank transaction data set, wherein the transaction graph comprises a plurality of account vertices, and the weight of the transaction edge between every two account vertices is obtained by the server 104 based on the matching degree assignment of the transaction feature between the corresponding two accounts and the preset abnormal feature. The server 104 carries out iterative calculation on the grade scores corresponding to the peaks of all accounts of the transaction graph until the grade scores are converged; the grade score corresponding to each account vertex is used for measuring the degree of abnormality of the account vertex relative to other account vertices in the transaction graph, and abnormal transaction results are identified from a plurality of account transactions corresponding to a plurality of account vertices based on the grade score corresponding to each account vertex when the grade score converges. The terminal 102 may be, but is not limited to, various personal computers, notebook computers, smart phones, tablet computers, and portable wearable devices, among others. The portable wearable device may be a smart watch, a headset, or the like. The server 104 may be implemented as a stand-alone server or as a server cluster of multiple servers.
In one embodiment, as shown in fig. 2, an abnormal transaction identification method is provided, and the method is applied to the server 104 in fig. 1 for illustration, and includes the following steps:
step S202, acquiring a bank transaction data set, and constructing a transaction diagram according to the bank transaction data set; the transaction graph comprises a plurality of account vertexes, and the weight of the transaction edge between every two account vertexes is obtained by assigning a value based on the matching degree of the transaction characteristic between the corresponding two accounts and the preset abnormal characteristic.
The bank transaction data set comprises transaction records among accounts, wherein the transaction records contain information such as transaction objects, transaction types, transaction amounts, transaction time and the like. Abnormal characteristics may refer to characteristics that are generally considered to exist for an abnormal transaction, such as an abnormal transaction amount, an abnormal transaction frequency, an abnormal transaction flow direction, and the like.
Optionally, the server acquires a data set of the bank transaction, including information such as transaction records and transaction time, and constructs a transaction diagram according to the transaction data, wherein the transaction diagram includes a plurality of account vertices, and the weight of the transaction edge between every two account vertices is obtained based on the matching degree assignment of the transaction feature and the abnormal feature between the corresponding two accounts. For example, a transaction graph G (V, E) is constructed according to the transaction data set, where V is the account set, E is the transaction edge between the account vertices, and the server assigns weights of different magnitudes to the transaction edge between the account vertices according to the matching degree between the transaction features and the preset abnormal features.
Step S204, performing iterative computation on the grade scores corresponding to the peaks of each account of the transaction graph until the grade scores are converged; the rank score corresponding to each account vertex is used to measure the degree of anomaly of that account vertex relative to other account vertices in the transaction map.
The iterative calculation refers to a process of continuously recursively calculating new values by using old values of variables until the error is smaller than a preset allowable error to complete the iterative calculation. The condition for convergence may be that the difference between the class fraction value and the last calculation is less than a certain set threshold, or that the maximum number of cycles is met.
Optionally, the server performs iterative computation on the grade score corresponding to each account vertex of the transaction graph until the grade score meets a preset convergence condition, where the grade score corresponding to each vertex is used to measure the degree of abnormality of the account vertex relative to other account vertices in the transaction graph, for example, the grade score corresponding to the account vertex is higher than the other account vertices, and the degree of abnormality of the account vertex is greater than the other account vertices.
Step S206, based on the grade scores corresponding to the vertexes of each account when the grade scores are converged, abnormal transaction results are identified from the plurality of account transactions corresponding to the vertexes of the accounts.
The abnormal transaction result may be a target account vertex screened out according to the grade score, and then a transaction with a larger transaction side weight is screened out from the related transactions as an abnormal transaction.
Optionally, the server sorts the account vertices exceeding the threshold value of the grade score according to the grade time sharing corresponding to each account vertex when the grade score converges, and identifies abnormal transactions from a plurality of transactions associated with the screened account vertices.
In the abnormal transaction identification method, by constructing the transaction graph comprising a plurality of account vertexes and transaction edges between the two account vertexes, different weights are given according to the matching degree of the transaction characteristics of the transaction edges and the preset abnormal characteristics, and the transaction edges with higher weights are more easily screened out when the grade scores corresponding to the account vertexes of the transaction graph are subjected to iterative calculation. Iterative calculation is carried out until the grade scores are converged, and based on the grade scores corresponding to the vertexes of each account when the grade scores are converged, abnormal transaction results are obtained through screening by identifying transaction edges corresponding to transaction points of the vertexes of the accounts with higher grade scores, so that the identification accuracy of the abnormal transactions in bank transactions is improved.
In one embodiment, step S204 performs iterative computation on the rank scores corresponding to the vertices of each account of the transaction map until the rank scores converge, and further includes:
determining total number N of account vertices in the transaction graph; the rank score for each account vertex is initialized to 1/N.
The account vertexes refer to accounts in a transaction diagram, and the transaction diagram comprises a plurality of account vertexes.
The grade score refers to a value obtained through calculation and is used for measuring the abnormal degree of the corresponding transaction corresponding to the corresponding account vertex. Initialization refers to setting the same initial value of 1/N for the rank scores of all the account vertices in the transaction graph.
Optionally, the server determines a total number N of account vertices included in the transaction map. The server sets the initial value of the rank scores of all the account vertices in the transaction map to 1/N.
In this embodiment, the same initial value of the grade score is set in the transaction diagram, so as to improve the operation speed of the subsequent iterative computation.
In one embodiment, step S202 obtains a banking transaction data set, and after constructing a transaction map according to the banking transaction data set, further includes:
step S202a, dividing the plurality of account vertices included in the transaction graph to obtain a plurality of disjoint account subsets.
The division processing refers to dividing a plurality of account vertices in the transaction graph according to the number of vertices. The disjoint may be that no duplicate accounts are included between the respective subsets of accounts.
Optionally, the server divides a plurality of account vertices included in the transaction map according to the number of the vertex points to obtain a plurality of disjoint account subsets. For example, the transaction diagram includes: the accounts corresponding to the six account vertexes of the account A, the account B, the account C, the account D, the account E and the account F are divided into two subsets of the accounts V1 = { A, B, C } and V2 = { D, E and F } in the transaction diagram.
Step S202b, constructing transaction subgraphs corresponding to each account subset; the transaction sub-graph includes account vertices for accounts in the subset of accounts.
The construction refers to reconstructing a small part of transaction edges among the missing account vertices caused by the segmentation processing in each account subset.
Optionally, the server constructs a corresponding transaction sub-graph for the subset of accounts; the transaction sub-graph includes account vertices for accounts in the subset of accounts and transaction edges between the accounts. For example, the subset of accounts is v1= { a, B, C }, and the corresponding subgraph g1= (V1, E1) is constructed from the subset: v1= { a, B, C }, e1= { a- > B, B- > C, C- > a }, where E1 is a transaction edge comprising the original transaction edge between account vertices and the reconstructed transaction edge.
In the embodiment, the account subset is obtained by dividing the account set in the transaction graph, the transaction subgraph is constructed based on the account subset, and iterative computation is guaranteed for the level score of the vertex of the account in parallel between each subsequent transaction subgraph, so that the computation processing efficiency is improved.
In one embodiment, step S202b further includes performing iterative computation on the rank scores corresponding to the vertices of each account of the transaction map until the rank scores converge, which specifically includes:
and carrying out iterative computation on the transaction subgraphs in parallel according to a preset PageRank algorithm until the grade scores corresponding to the account vertexes of the transaction subgraphs are converged.
Parallel processing means parallel processing, namely, iterative computation is carried out on a plurality of transaction subgraphs at the same time.
Optionally, the server performs iterative computation on the transaction subgraphs at the same time according to a preset PageRank algorithm until the grade scores corresponding to the account vertices of the transaction subgraphs converge.
In this embodiment, the effect of improving the calculation efficiency is achieved by performing iterative calculation according to the preset PageRank algorithm on a plurality of transaction subgraphs at the same time.
In one embodiment, as shown in fig. 3, according to a preset PageRank algorithm, iterative computation is performed on multiple transaction subgraphs in parallel until the rank scores corresponding to the account vertices of each transaction subgraph converge, including:
Step S302, in the iterative calculation process of the grade score of any account vertex in each transaction sub-graph, the weight of each transaction edge corresponding to any account vertex is obtained, and the weights of all transaction edges corresponding to any account vertex are added to obtain the sum of the weights of the transaction edges corresponding to any account vertex.
The weight is a quantification concept relative to the overall evaluation and is used for measuring the relative importance of a certain factor in the overall evaluation. In the process of comprehensive evaluation, decision making and the like, the weight is a relative importance coefficient of distribution of different evaluation factors or influence factors.
Optionally, in the iterative calculation process of the grade scores of all the account vertices in each transaction sub-graph, the server obtains the weight of each transaction edge corresponding to each account vertex, and adds the weights of all the transaction edges corresponding to each account vertex to obtain the sum of the weights of the transaction edges corresponding to each account vertex.
Step S304, calculating the grade score corresponding to any account vertex based on a PageRank algorithm preset by the weight of each transaction edge of any account vertex and the sum of the weights of the transaction edges, and obtaining the grade score updated by any account vertex.
The calculating may be performed by substituting the sum of the weight of each transaction edge of any account vertex and the weight of the transaction edge into a preset PageRank algorithm formula.
Optionally, the server substitutes the sum of the weight of each transaction edge of each account vertex and the weight of each transaction edge into a preset PageRank algorithm formula, and calculates the grade score corresponding to each account vertex to obtain the grade score corresponding to each account vertex after being updated.
In this embodiment, the weight of the transaction edge corresponding to each account vertex is obtained by adding the weights of all the transaction edges of each account vertex, and then the weights are substituted into a formula to calculate the corresponding grade score after updating each account vertex, so that the grade score is calculated in each iteration.
In one embodiment, step S302 calculates a rank score corresponding to any account vertex based on a PageRank algorithm preset by a sum of a weight of each transaction edge and a weight of each transaction edge of any account vertex, to obtain a rank score updated by any account vertex, including:
and obtaining the current grade score of any account vertex. And updating and calculating the grade score corresponding to any account vertex based on the current grade score of any account vertex, a preset damping coefficient, the weight of each transaction edge of any account vertex and the weight sum of the transaction edges, so as to obtain the updated grade score.
The current grade score refers to a grade score value obtained by the last iterative calculation.
The damping coefficient may be a probability that the account is engaged and the transaction is completed. The formula for performing the calculation substitution may be:
PR(u)＝(1-d)/N+d*∑(PR(v')*w(u,v')/w_sum)
where v ' is the neighbor vertex of u, w (u, v ') is the link weight between vertices u and v ', w_sum is the link weight sum of vertices u, and d is the damping coefficient. The 1/N may be the initial grade score of the vertex of the account, and when the iterative calculation is performed, the grade score obtained by the previous iterative calculation is obtained.
Optionally, the server acquires a grade score value obtained by last iterative computation of each account vertex, and if the calculation is performed for the first time, acquires an initial grade score 1/N of the account vertex. The server substitutes the current grade score of each account vertex, a preset damping coefficient, the weight of all corresponding transaction edges of each account vertex and the sum of the weights of all transaction edges into a formula of a PageRank algorithm to calculate, and the grade score of each account vertex after updating is calculated.
In this embodiment, the current grade score of each account vertex is obtained, and then the current grade score is substituted into a calculation formula with other parameters obtained before to calculate, so as to obtain the updated grade score of the account vertex, thereby realizing iterative calculation of the grade score of the account vertex.
In one embodiment, as shown in fig. 4, the step S302 of obtaining the weight of each transaction edge corresponding to any account vertex includes:
step S402, obtaining the current time as the first time, and obtaining the transaction occurrence time corresponding to each transaction edge of any account vertex as the second time.
The current time refers to the time when the grade score calculation is needed to be carried out on the vertex of the account each time.
Optionally, when the class score calculation is required for the account vertices, the server obtains the current time as a first time, and obtains the time of occurrence of the transaction corresponding to each account vertex and each associated transaction edge in the transaction sub-graph as a second time.
Step S404, based on the first time and the second time, obtaining a time attenuation factor corresponding to each transaction edge of any account vertex.
The time attenuation factor can be calculated by the following formula:
decay_factor＝exp(-α*(t-t'))
where α is a time decay coefficient for controlling the decay rate. A larger value of α means that the decay rate of the transaction is faster, a smaller value of α means that the decay rate is slower, the decay factor is the time decay factor, t is the current time, and t' is the time at which the transaction occurs.
Optionally, the server substitutes the first time and the second time into a calculation formula of the time attenuation factor, and the time attenuation factor corresponding to each transaction edge of each account vertex is obtained through calculation. For example, assuming there is a transaction diagram between three accounts, the transaction occurs for the following time:
account a- > account B:2023-05-01 10:00:00
Account B- > account C:2023-05-02 15:30:00
Account C- > account a:2023-05-03 12:45:00
The current time is 2023-06-01 14:00:00, assuming that the time decay coefficient α is 0.01.
According to a calculation formula of the time attenuation factor, we can calculate the attenuation factor value of each transaction: decay_factor_a- > b=exp (-0.01 x (2023-06-01:00:00-2023-05-01:10:00:00));
decay_factor_B->C＝exp(-0.01*(2023-06-01 14:00:00-2023-05-02 15:30:00))；decay_factor_C->A＝exp(-0.01*(2023-06-01 14:00:00-2023-05-03 12:45:00))。
the calculation results are assumed to be as follows:
decay_factor_A->B＝0.85；
decay_factor_B->C＝0.95；
decay_factor_C->A＝0.90。
step S406, multiplying the time attenuation factor by the weight of the corresponding transaction edge in the last iteration calculation to obtain the weight of the adjusted transaction edge, and taking the weight of the transaction edge in the iteration calculation.
The adjusted weight of the transaction edge is used as the weight of the transaction edge in the iterative calculation to be substituted into a calculation formula for calculation.
Optionally, the server multiplies the calculated time attenuation factor by the weight of the corresponding transaction edge in the last iteration calculation to obtain an adjusted transaction edge weight, and the adjusted transaction edge weight is used as the transaction edge weight in the current iteration calculation.
In this embodiment, the time attenuation factor is introduced to adjust the weight of each transaction edge, and then the adjusted weight is used as the transaction edge weight in the iterative calculation, so that the influence of the transaction before the transaction time is longer on the whole abnormal transaction identification process is reduced, and the attention degree of the transaction to the latest transaction in the abnormal transaction identification is improved.
In one embodiment, based on all the embodiments above, the transaction characteristics between the two accounts include transaction edges including transaction amount, transaction frequency, and transaction flow direction; the preset abnormal characteristics comprise abnormal amount characteristics, abnormal frequency characteristics and abnormal flow direction characteristics.
The method further comprises the following steps:
the method comprises the steps of obtaining a first matching degree of transaction amount and abnormal amount characteristics between two corresponding accounts, a second matching degree of transaction frequency and abnormal transaction frequency between two corresponding accounts, and a third matching degree of transaction flow direction and abnormal flow direction characteristics between two corresponding accounts. And determining the weight of the transaction edge between the two corresponding accounts based on the first matching degree, the second matching degree and the third matching degree.
Wherein the abnormal amount feature may be a transaction amount with mantissas of 8, 9 or an end of integer transaction amount. The abnormal transaction frequency may be a higher transaction frequency. The abnormal flow direction characteristic may be a concentrated sink and a scattered sink, a scattered sink and a concentrated sink.
Optionally, the server obtains the transaction amount, the transaction frequency and the transaction flow direction between the two accounts, and matches the transaction amount, the transaction frequency and the transaction flow direction with the abnormal amount characteristic, the abnormal transaction frequency and the abnormal flow direction characteristic respectively, and the higher the matching degree is, the larger the set weight is. For example, matching transaction amount with abnormal amount characteristics: when 8,9 is mantissa in the transaction amount, the weight is set to 1.5; when the integer in the transaction amount is mantissa, the weight is set to be 1.5; when the transaction amount is the rest mantissas, the weight is set to be 1; matching the transaction frequency with an abnormal transaction frequency: when the transaction frequency is 0-10 times, setting the weight as 1; when the transaction frequency is 10-100 times, the weight is set to be 1.2; when the transaction frequency is greater than 100 times, the weight is set to 1.5; matching the transaction flow direction with the abnormal flow direction characteristics: when the inflow/outflow number >20, the weight is set to 1.5; when the number of outflow/inflow >20, the weight is set to 1.5; the remaining proportion sets the weight to 1.
In this embodiment, by further defining transaction characteristics among accounts, matching the transaction characteristics with preset abnormal transaction characteristics, and setting different weights for corresponding transaction edges according to the matching degree, the recognition accuracy of abnormal transactions is improved.
In one embodiment, as shown in fig. 5, there is provided a fraud transaction detection method based on an optimization graph algorithm, which specifically includes the following steps:
step S502, acquiring a bank transaction data set, and constructing a transaction diagram according to the bank transaction data set; the transaction graph comprises a plurality of account vertexes, and the weight of the transaction edge between every two account vertexes is obtained by assigning a value based on the matching degree of the transaction characteristic between the corresponding two accounts and the preset abnormal characteristic.
Specifically, the server acquires a data set of bank transaction, including information such as transaction records and transaction time, and constructs a transaction diagram according to the transaction data, wherein the transaction diagram comprises a plurality of account vertices, and the weight of a transaction edge between every two account vertices is obtained based on matching degree assignment of transaction characteristics and abnormal characteristics between two corresponding accounts. For example, a transaction graph G (V, E) is constructed according to the transaction data set, where V is the account set, E is the transaction edge between the account vertices, and the server assigns weights of different magnitudes to the transaction edge between the account vertices according to the matching degree between the transaction features and the preset abnormal features.
Step S504, the transaction characteristics between the two accounts comprise transaction edges including transaction amount, transaction frequency and transaction flow direction; the preset abnormal characteristics comprise abnormal amount characteristics, abnormal frequency characteristics and abnormal flow direction characteristics; further comprises: the method comprises the steps of obtaining a first matching degree of transaction amount and abnormal amount characteristics between two corresponding accounts, a second matching degree of transaction frequency and abnormal transaction frequency between two corresponding accounts, and a third matching degree of transaction flow direction and abnormal flow direction characteristics between two corresponding accounts. And determining the weight of the transaction edge between the two corresponding accounts based on the first matching degree, the second matching degree and the third matching degree.
Specifically, the server acquires the transaction amount, the transaction frequency and the transaction flow direction between the two accounts, and respectively matches the transaction amount, the transaction frequency and the transaction flow direction with the abnormal amount characteristic, the abnormal transaction frequency and the abnormal flow direction characteristic, and the higher the matching degree is, the larger the set weight is.
Step S506, dividing a plurality of account vertices included in the transaction graph to obtain a plurality of disjoint account subsets; constructing a transaction subgraph corresponding to each account subset; the transaction sub-graph includes account vertices for accounts in the subset of accounts.
Specifically, the server divides a plurality of account vertices included in the transaction map according to the number of the vertex points to obtain a plurality of disjoint account subsets. The server constructs a corresponding transaction sub-graph for the account sub-set; the transaction sub-graph includes account vertices for accounts in the subset of accounts and transaction edges between the accounts.
Step S508, performing iterative computation on the transaction subgraphs in parallel according to a preset PageRank algorithm until the grade scores corresponding to the account vertices of the transaction subgraphs are converged.
Specifically, the server performs iterative computation on the transaction subgraphs at the same time according to a preset PageRank algorithm until the grade scores corresponding to the account vertices of the transaction subgraphs are converged.
Step S510, obtaining the current time as the first time, and obtaining the transaction occurrence time corresponding to each transaction edge of any account vertex as the second time; based on the first time and the second time, obtaining a time attenuation factor corresponding to each transaction edge of any account vertex; and multiplying the time attenuation factor by the weight of the corresponding transaction edge in the last iteration calculation to obtain the weight of the adjusted transaction edge, wherein the weight of the transaction edge in the iteration calculation is used as the weight of the transaction edge.
Specifically, when the grade score calculation is required to be performed on the account vertices, the server acquires the current time as the first time, and acquires the time of occurrence of the transaction corresponding to each account vertex and each associated transaction edge in the transaction sub-graph as the second time. And substituting the first time and the second time into a calculation formula of the time attenuation factor by the server, and obtaining the time attenuation factor corresponding to each transaction edge of each account vertex through calculation. The server multiplies the calculated time attenuation factor by the weight of the corresponding transaction edge in the last iteration calculation to obtain the adjusted transaction edge weight, and the adjusted transaction edge weight is used as the transaction edge weight in the iteration calculation.
Step S512, in the iterative calculation process of the grade score of any account vertex in each transaction sub-graph, the weight of each transaction edge corresponding to any account vertex is obtained, and the weights of all transaction edges corresponding to any account vertex are added to obtain the sum of the weights of the transaction edges corresponding to any account vertex.
Specifically, in the iterative calculation process of the grade scores of all the account vertices in each transaction sub-graph, the server obtains the weight of each transaction edge corresponding to each account vertex, and adds the weights of all the transaction edges corresponding to each account vertex to obtain the sum of the weights of the transaction edges corresponding to each account vertex.
Step S514, calculating the grade score corresponding to any account vertex based on the weight of each transaction edge of any account vertex, the weight of the transaction edge, the preset damping coefficient and the preset PageRank algorithm, and obtaining the grade score updated by any account vertex.
Specifically, the server substitutes the sum of the weight of each transaction edge of each account vertex and the weight of each transaction edge into a preset PageRank algorithm formula, calculates the grade score corresponding to each account vertex, and obtains the grade score corresponding to each account vertex after being updated.
Step S516, based on the grade scores corresponding to the vertexes of each account when the grade scores are converged, abnormal transaction results are identified from the plurality of account transactions corresponding to the vertexes of the accounts.
Specifically, the server sorts the account vertices exceeding the threshold value of the grade score according to the grade time sharing corresponding to each account vertex when the grade score converges, screens out the account vertices exceeding the threshold value of the grade score, and identifies abnormal transactions from a plurality of transactions associated with the screened account vertices.
In this embodiment, by constructing a transaction graph including a plurality of account vertices and transaction edges between two account vertices, different weights are given according to matching degrees of transaction features of the transaction edges and preset abnormal features, and the transaction edges with higher weights are more easily screened out when the grade scores corresponding to the account vertices of the transaction graph are calculated in an iterative manner according to a preset PageRank algorithm. Iterative calculation is carried out until the grade scores are converged, and based on the grade scores corresponding to the vertexes of each account when the grade scores are converged, abnormal transaction results are obtained through screening by identifying transaction edges corresponding to the vertexes of the accounts with higher grade scores, so that the accuracy of abnormal transaction identification in bank transactions is improved.
It should be understood that, although the steps in the flowcharts related to the embodiments described above are sequentially shown as indicated by arrows, these steps are not necessarily sequentially performed in the order indicated by the arrows. The steps are not strictly limited to the order of execution unless explicitly recited herein, and the steps may be executed in other orders. Moreover, at least some of the steps in the flowcharts described in the above embodiments may include a plurality of steps or a plurality of stages, which are not necessarily performed at the same time, but may be performed at different times, and the order of the steps or stages is not necessarily performed sequentially, but may be performed alternately or alternately with at least some of the other steps or stages.
Based on the same inventive concept, the embodiment of the application also provides an abnormal transaction identification device for realizing the abnormal transaction identification method. The implementation of the solution provided by the device is similar to the implementation described in the above method, so the specific limitation in the embodiment of the abnormal transaction recognition device or devices provided below may refer to the limitation of the abnormal transaction recognition method hereinabove, and will not be repeated herein.
In one embodiment, as shown in FIG. 6, there is provided an abnormal transaction identification apparatus 600 comprising: a graph construction module 602, a score calculation module 604, and an anomaly determination module 606, wherein:
the diagram construction module 602 is configured to acquire a banking transaction data set, and construct a transaction diagram according to the banking transaction data set; the transaction graph comprises a plurality of account vertexes, and the weight of the transaction edge between every two account vertexes is obtained by assigning a value based on the matching degree of the transaction characteristic between the corresponding two accounts and the preset abnormal characteristic.
The score calculating module 604 is configured to iteratively calculate a grade score corresponding to each account vertex of the transaction map according to a preset PageRank algorithm until the grade score converges; the rank score corresponding to each account vertex is used to measure the degree of anomaly of that account vertex relative to other account vertices in the transaction map.
The anomaly determination module 606 is configured to identify an anomaly transaction result from a plurality of account transactions corresponding to the plurality of account vertices based on the rank score corresponding to each account vertex when the rank scores converge.
Further, in another embodiment, the apparatus 600 further includes an initialization module, configured to iteratively calculate the rank scores corresponding to the account vertices of the transaction graph according to a preset PageRank algorithm, until the rank scores converge, and determine the total number N of the account vertices in the transaction graph; the rank score for each account vertex is initialized to 1/N.
Further, in another embodiment, the graph construction module 602 is further configured to obtain a banking transaction data set, and after constructing a transaction graph according to the banking transaction data set, segment a plurality of account vertices included in the transaction graph to obtain a plurality of disjoint account subsets; constructing a transaction subgraph corresponding to each account subset; the transaction sub-graph includes account vertices for accounts in the subset of accounts.
Further, in another embodiment, the score calculating module 604 is further configured to perform iterative calculation on the plurality of transaction subgraphs in parallel according to a preset PageRank algorithm until the rank scores corresponding to the account vertices of the transaction subgraphs converge.
Further, in another embodiment, the score calculating module 604 is further configured to obtain the weight of each transaction edge corresponding to any account vertex in the iterative calculation process of the grade score of any account vertex in each transaction sub-graph, and add the weights of all the transaction edges corresponding to any account vertex to obtain the sum of the weights of the transaction edges corresponding to any account vertex; and calculating the grade score corresponding to any account vertex based on a PageRank algorithm preset by the weight of each transaction edge of any account vertex and the sum of the weights of the transaction edges, so as to obtain the grade score updated by any account vertex.
Further, in another embodiment, the score calculating module 604 is further configured to obtain a current class score of any of the account vertices; and updating and calculating the grade score corresponding to any account vertex based on the current grade score of any account vertex, a preset damping coefficient, the weight of each transaction edge of any account vertex and the weight sum of the transaction edges, so as to obtain the updated grade score.
Further, in another embodiment, the score calculating module 604 is further configured to obtain the current time as a first time, and obtain the transaction occurrence time corresponding to each transaction edge of any vertex of the account as a second time; based on the first time and the second time, obtaining a time attenuation factor corresponding to each transaction edge of any account vertex; and multiplying the time attenuation factor by the weight of the corresponding transaction edge in the last iteration calculation to obtain the weight of the adjusted transaction edge, wherein the weight of the transaction edge in the iteration calculation is used as the weight of the transaction edge.
Further, in another embodiment, the transaction characteristics between the two accounts include transaction edges including transaction amount, transaction frequency, and transaction flow direction; the preset abnormal characteristics comprise abnormal amount characteristics, abnormal frequency characteristics and abnormal flow direction characteristics; the device further comprises a weight setting module, a weight setting module and a weight setting module, wherein the weight setting module is used for acquiring a first matching degree of the characteristics of the transaction amount and the abnormal amount between the two accounts, a second matching degree of the transaction frequency and the abnormal transaction frequency between the two accounts and a third matching degree of the transaction flow direction and the abnormal flow direction between the two accounts; and determining the weight of the transaction edge between the two corresponding accounts based on the first matching degree, the second matching degree and the third matching degree.
The respective modules in the abnormal transaction recognition apparatus described above may be implemented in whole or in part by software, hardware, and a combination thereof. The above modules may be embedded in hardware or may be independent of a processor in the computer device, or may be stored in software in a memory in the computer device, so that the processor may call and execute operations corresponding to the above modules.
In one embodiment, a computer device is provided, which may be a server, the internal structure of which may be as shown in fig. 7. The computer device includes a processor, a memory, an Input/Output interface (I/O) and a communication interface. The processor, the memory and the input/output interface are connected through a system bus, and the communication interface is connected to the system bus through the input/output interface. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The database of the computer device is used for storing banking transaction information data. The input/output interface of the computer device is used to exchange information between the processor and the external device. The communication interface of the computer device is used for communicating with an external terminal through a network connection. The computer program, when executed by a processor, implements a method of abnormal transaction identification.
It will be appreciated by those skilled in the art that the structure shown in fig. 7 is merely a block diagram of some of the structures associated with the present application and is not limiting of the computer device to which the present application may be applied, and that a particular computer device may include more or fewer components than shown, or may combine certain components, or have a different arrangement of components.
In an embodiment, there is also provided a computer device comprising a memory and a processor, the memory having stored therein a computer program, the processor implementing the steps of the method embodiments described above when the computer program is executed.
In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored which, when executed by a processor, carries out the steps of the method embodiments described above.
In an embodiment, a computer program product is provided, comprising a computer program which, when executed by a processor, implements the steps of the method embodiments described above.
It should be noted that, the user information (including, but not limited to, user equipment information, user personal information, etc.) and the data (including, but not limited to, data for analysis, stored data, presented data, etc.) referred to in the present application are information and data authorized by the user or sufficiently authorized by each party, and the collection, use and processing of the related data are required to comply with the related laws and regulations and standards of the related countries and regions.
Those skilled in the art will appreciate that implementing all or part of the above described methods may be accomplished by way of a computer program stored on a non-transitory computer readable storage medium, which when executed, may comprise the steps of the embodiments of the methods described above. Any reference to memory, database, or other medium used in the various embodiments provided herein may include at least one of non-volatile and volatile memory. The nonvolatile Memory may include Read-Only Memory (ROM), magnetic tape, floppy disk, flash Memory, optical Memory, high density embedded nonvolatile Memory, resistive random access Memory (ReRAM), magnetic random access Memory (Magnetoresistive Random Access Memory, MRAM), ferroelectric Memory (Ferroelectric Random Access Memory, FRAM), phase change Memory (Phase Change Memory, PCM), graphene Memory, and the like. Volatile memory can include random access memory (Random Access Memory, RAM) or external cache memory, and the like. By way of illustration, and not limitation, RAM can be in the form of a variety of forms, such as static random access memory (Static Random Access Memory, SRAM) or dynamic random access memory (Dynamic Random Access Memory, DRAM), and the like. The databases referred to in the various embodiments provided herein may include at least one of relational databases and non-relational databases. The non-relational database may include, but is not limited to, a blockchain-based distributed database, and the like. The processors referred to in the embodiments provided herein may be general purpose processors, central processing units, graphics processors, digital signal processors, programmable logic units, quantum computing-based data processing logic units, etc., without being limited thereto.
The technical features of the above embodiments may be arbitrarily combined, and all possible combinations of the technical features in the above embodiments are not described for brevity of description, however, as long as there is no contradiction between the combinations of the technical features, they should be considered as the scope of the description.
The above examples only represent a few embodiments of the present application, which are described in more detail and are not to be construed as limiting the scope of the present application. It should be noted that it would be apparent to those skilled in the art that various modifications and improvements could be made without departing from the spirit of the present application, which would be within the scope of the present application. Accordingly, the scope of protection of the present application shall be subject to the appended claims.

Claims (12)

1. A method of identifying an abnormal transaction, the method comprising:
acquiring a banking transaction data set, and constructing a transaction diagram according to the banking transaction data set; the transaction graph comprises a plurality of account vertexes, and the weight of the transaction edge between every two account vertexes is obtained by assigning a value based on the matching degree of the transaction characteristic between the corresponding two accounts and the preset abnormal characteristic;
Performing iterative computation on the grade scores corresponding to the peaks of all accounts of the transaction graph until the grade scores are converged; the grade score corresponding to each account vertex is used for measuring the degree of abnormality of the account vertex relative to other account vertices in the transaction diagram;
and identifying abnormal transaction results from the plurality of account transactions corresponding to the plurality of account vertices based on the grade scores corresponding to each account vertex when the grade scores are converged.
2. The method of claim 1, wherein iteratively calculating the rank scores for the account vertices of the transaction map until the rank scores converge, further comprises:
determining total number N of account vertices in the transaction graph;
initializing the grade score of each account vertex to be 1/N.
3. The method of claim 1, wherein after the acquiring the banking data set and constructing the transaction map from the banking data set, further comprises:
dividing a plurality of account vertices included in the transaction graph to obtain a plurality of disjoint account subsets;
constructing transaction subgraphs corresponding to each account subset; the transaction sub-graph includes account vertices for accounts in the subset of accounts.
4. The method of claim 3, wherein iteratively calculating the rank scores for the account vertices of the transaction map until the rank scores converge comprises:
and carrying out iterative computation on a plurality of transaction subgraphs in parallel according to a preset PageRank algorithm until the grade scores corresponding to the account vertexes of the transaction subgraphs are converged.
5. The method according to claim 4, wherein the iterative computation of the transaction sub-graphs in parallel according to the preset PageRank algorithm until the rank score corresponding to the account vertex of each transaction sub-graph converges, comprises:
in the iterative calculation process of the grade score of any account vertex in each transaction sub-graph, obtaining the weight of each transaction edge corresponding to any account vertex, and adding the weights of all transaction edges corresponding to any account vertex to obtain the sum of the weights of the transaction edges corresponding to any account vertex;
and calculating the grade score corresponding to any account vertex based on a PageRank algorithm preset by the weight of each transaction edge of any account vertex and the sum of the weights of the transaction edges, so as to obtain the grade score updated by any account vertex.
6. The method of claim 5, wherein the calculating the rank score corresponding to the any account vertex based on a PageRank algorithm preset by a sum of a weight of each transaction edge of the any account vertex and a weight of the transaction edge to obtain the rank score updated by the any account vertex comprises:
obtaining the current grade score of any account vertex;
and updating and calculating the grade score corresponding to any account vertex based on the current grade score of any account vertex, the preset damping coefficient, the weight of each transaction edge of any account vertex and the sum of the weights of the transaction edges, so as to obtain the updated grade score.
7. The method of claim 5, wherein obtaining the weight of each transaction edge corresponding to the vertex of any account comprises:
acquiring the current time as a first time, and acquiring the transaction occurrence time corresponding to each transaction edge of any account vertex as a second time;
obtaining a time attenuation factor corresponding to each transaction edge of any account vertex based on the first time and the second time;
And multiplying the time attenuation factor by the weight of the corresponding transaction edge in the last iteration calculation to obtain the weight of the adjusted transaction edge, wherein the weight of the transaction edge in the iteration calculation is used as the weight of the transaction edge.
8. The method of any one of claims 1 to 7, wherein the transaction characteristics between the two accounts include transaction edges including transaction amount, transaction frequency, and transaction flow direction; the preset abnormal characteristics comprise abnormal amount characteristics, abnormal frequency characteristics and abnormal flow direction characteristics;
the method further comprises the steps of:
acquiring a first matching degree of transaction amount between two corresponding accounts and the characteristic of the abnormal amount, a second matching degree of transaction frequency between two corresponding accounts and the abnormal transaction frequency, and a third matching degree of transaction flow direction between two corresponding accounts and the characteristic of the abnormal flow direction;
and determining the weight of the transaction edge between the two corresponding accounts based on the first matching degree, the second matching degree and the third matching degree.
9. An abnormal transaction identification device, the device comprising:
the diagram construction module is used for acquiring a bank transaction data set and constructing a transaction diagram according to the bank transaction data set; the transaction graph comprises a plurality of account vertexes, and the weight of the transaction edge between every two account vertexes is obtained by assigning a value based on the matching degree of the transaction characteristic between the corresponding two accounts and the preset abnormal characteristic;
The score calculation module is used for carrying out iterative calculation on the grade scores corresponding to the vertexes of each account of the transaction graph until the grade scores are converged; the grade score corresponding to each account vertex is used for measuring the degree of abnormality of the account vertex relative to other account vertices in the transaction diagram;
and the abnormality judgment module is used for identifying abnormal transaction results from the plurality of account transactions corresponding to the plurality of account vertices based on the grade scores corresponding to the account vertices when the grade scores are converged.
10. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that the processor implements the steps of the method of any one of claims 1 to 8 when the computer program is executed.
11. A computer readable storage medium, on which a computer program is stored, characterized in that the computer program, when being executed by a processor, implements the steps of the method of any of claims 1 to 8.
12. A computer program product comprising a computer program, characterized in that the computer program, when executed by a processor, implements the steps of the method of any one of claims 1 to 8.
CN202310883373.7A 2023-07-18 2023-07-18 Abnormal transaction identification method, device, computer equipment and storage medium Pending CN117436882A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202310883373.7A CN117436882A (en) 2023-07-18 2023-07-18 Abnormal transaction identification method, device, computer equipment and storage medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202310883373.7A CN117436882A (en) 2023-07-18 2023-07-18 Abnormal transaction identification method, device, computer equipment and storage medium

Publications (1)

Publication Number Publication Date
CN117436882A true CN117436882A (en) 2024-01-23

Family

ID=89545158

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202310883373.7A Pending CN117436882A (en) 2023-07-18 2023-07-18 Abnormal transaction identification method, device, computer equipment and storage medium

Country Status (1)

Country Link
CN (1) CN117436882A (en)

Similar Documents

Publication Publication Date Title
TWI662421B (en) Community division method and device based on feature matching network
US20190340533A1 (en) Systems and methods for preparing data for use by machine learning algorithms
US20210081798A1 (en) Neural network method and apparatus
WO2023103527A1 (en) Access frequency prediction method and device
CN114187112A (en) Training method of account risk model and determination method of risk user group
CN115943397A (en) Federated dual random core learning of vertically partitioned data
US20240126730A1 (en) Schema Validation with Data Synthesis
US20160253674A1 (en) Efficient tail calculation to exploit data correlation
CN117436882A (en) Abnormal transaction identification method, device, computer equipment and storage medium
US20230325757A1 (en) Computer-implemented method and system for testing a model
CN112347425B (en) Method and system for dense subgraph detection based on time sequence
CN114238044A (en) Method and device for calculating activity of open source project and computer equipment
CN113159937A (en) Method and device for identifying risks and electronic equipment
CN114080609A (en) Nonlinear causal modeling based on coding knowledge
CN113344587B (en) Data grade determining method and device, electronic equipment and storage medium
US10719490B1 (en) Forensic analysis using synthetic datasets
CN117726431A (en) Credit information updating method, apparatus, device, storage medium and program product
US11487765B1 (en) Generating relaxed synthetic data using adaptive projection
US20240104421A1 (en) Correlation-based dimensional reduction of synthesized features
US20230419098A1 (en) Utilizing selective transformation and replacement with high-dimensionality projection layers to implement neural networks in tabular data environments
US20240104436A1 (en) Chained feature synthesis and dimensional reduction
CN117522138A (en) Method, device, equipment and medium for identifying testing risk of financial business system
CN114218397A (en) Transaction relation map processing method and device, computer equipment and storage medium
CN117275007A (en) Account information recognition classifier generation method and device and resource card recognition method
WO2023215752A1 (en) Training a neural network model across multiple domains

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination