CN110135856A - A kind of repeat business risk monitoring method, device and computer readable storage medium - Google Patents
A kind of repeat business risk monitoring method, device and computer readable storage medium Download PDFInfo
- Publication number
- CN110135856A CN110135856A CN201910405528.XA CN201910405528A CN110135856A CN 110135856 A CN110135856 A CN 110135856A CN 201910405528 A CN201910405528 A CN 201910405528A CN 110135856 A CN110135856 A CN 110135856A
- Authority
- CN
- China
- Prior art keywords
- similarity
- batch
- message
- parameter
- measured
- 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.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Accounting & Taxation (AREA)
- Physics & Mathematics (AREA)
- Strategic Management (AREA)
- Bioinformatics & Computational Biology (AREA)
- Finance (AREA)
- Computer Security & Cryptography (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Cheminformatics (AREA)
- General Business, Economics & Management (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Debugging And Monitoring (AREA)
- Financial Or Insurance-Related Operations Such As Payment And Settlement (AREA)
Abstract
The present invention provides a kind of repeat business risk monitoring method, device and computer readable storage mediums, method include: obtain the batch transaction message to be measured sent in given time by same monitoring object, and before given time on the historical trading message that send；According to specified message content, the index similarity between batch transaction message to be measured and historical trading message is determined；By being compared to index similarity with default similarity threshold, to judge batch transaction message to be measured with the presence or absence of repeat business risk.It, can be to being monitored the case where part repeat business that may be present in the batch transaction message sent on different batches using the above method, and then repeat business risk more can be delicately prompted, avoid economic loss.
Description
Technical field
The invention belongs to trading processing technical fields, and in particular to a kind of repeat business risk monitoring method, device and meter
Calculation machine readable storage medium storing program for executing.
Background technique
Background that this section is intended to provide an explanation of the embodiments of the present invention set forth in the claims or context.Herein
Description recognizes it is the prior art not because not being included in this section.
In financial field, " writing instructions and transfer real " is a kind of common approach for handling batch service, typically refers to batch to be processed
Amount transaction is converted into a kind of technology that real-time deal is handled.For example, using batch text between accepting institution and Unionpay's system
Part mode transmits transaction message, and transmits transaction message using online message mode between Unionpay's system and card sending mechanism.However,
When abnormal due to server resource deficiency, network environment Caton or service end system shake etc., in fact it could happen that transaction repeats hair
The problem of sending, and then lead to economic loss.
In order to solve the problems, such as the repeat business occurred in above-mentioned batch transaction, the method generallyd use in the prior art is logical
The batch number for comparing the batch transaction message being currently received between the batch transaction message that receives before is crossed, to receive for the first time
Subject to the batch number arrived, and the duplicate batch transaction message of batch number is abandoned；However, being only with batch number in above scheme
Discrimination standard, and it is not directed to the specifying information of transaction, if repeat business occurs in the file of different batches, existing scheme
It will be unable to identify.
Summary of the invention
It is difficult in the prior art for above-mentioned to the part repeat business in the batch transaction message for being present in different batches
It is monitored this problem, proposes a kind of repeat business risk monitoring method, device, system and computer-readable storage medium
Matter is able to solve the above problem with this method, device, system and computer readable storage medium.
The present invention provides following scheme.
A kind of repeat business risk monitoring method, include: obtain sent in given time by same monitoring object it is to be measured
Batch transaction message, and before given time on the historical trading message that send；According to specified message content, determine to be measured
Index similarity between batch transaction message and historical trading message, wherein specified message content include in following at least
Two kinds: batch number, Transaction Account number and transaction amount；By being compared to index similarity with default similarity threshold, with
Judge batch transaction message to be measured with the presence or absence of repeat business risk.
In some possible embodiments, it obtains and is traded by the batch to be measured that same monitoring object is sent in given time
Message, and before given time on the historical trading message that send include: to receive in given time by same monitoring object
The batch transaction message to be measured sent；First period was determined by preset duration and given time, and extracts same monitoring object the
The historical trading message sent in one period.
In some possible embodiments, the similarity between batch transaction message to be measured and historical trading message is determined
Index include: using default similarity algorithm determine similarity between batch transaction message to be measured and historical trading message to
Amount；Using default code of points, index similarity is converted by similarity vector.
In some possible embodiments, determine that batch transaction message to be measured and history are handed over using default similarity algorithm
Similarity vector between easy message includes: to construct sparse matrix based on batch transaction message to be measured and historical trading message, dilute
Dredge in matrix, the value of each nonzero element is determined by transaction amount, the row label of each element and column label respectively by
Batch number and Transaction Account number determine；Determine m phase between the first sparse vector and m the second sparse vectors in sparse matrix
Like degree parameter, and similarity vector is determined by m similarity parameter；Wherein, batch transaction message to be measured includes: corresponding to first
More transaction messages of batch number, row vector/or the column vector that first batch number is corresponded in sparse matrix are sparse as first
Vector；Historical trading message includes: the more transaction messages for corresponding respectively to m second lot number, right respectively in sparse matrix
Should in m second lot number row vector/or column vector as m the second sparse vectors, m is positive integer.
In some possible embodiments, further includes: by # { (bi- a) ≠ 0 and # { (bi+ a) ≠ 0 ratio and/or
Difference determines m similarity parameter between the first sparse vector in sparse matrix and m the second sparse vectors, wherein i=
1,2,...,m；Wherein, biIndicate i-th of second sparse vectors in m the second sparse vectors, a indicates the first sparse vector, #
{(bi- a) ≠ 0 indicate the number of nonzero element in the difference vector of the first sparse vector and i-th of second sparse vectors, # { (b+
A) ≠ 0 } indicate the first sparse vector and i-th second sparse vectors and in vector nonzero element number.
Specifically: the m between the first sparse vector in sparse matrix and m the second sparse vectors is determined by following equation
A similarity parameter:
In some possible embodiments, default code of points comprises determining that the maximum in m similarity parameter is similar
Parameter is spent as index similarity.
In some possible embodiments, code of points is preset further include: judge the maximum phase in m similarity parameter
Whether reach preset critical like degree parameter；If maximum similarity parameter reaches preset critical, it is determined that preset critical is
Index similarity；If maximum similarity parameter is not up to preset critical, based on m preset weights parameter respectively to m phase
It is weighted processing like degree parameter, to obtain m Weighted Similarity parameter, and determines the maximum in m Weighted Similarity parameter
Weighted Similarity parameter is as index similarity.
In some possible embodiments, specified message content further includes that the time is sent in batch, method further include: be directed to
Each of m similarity parameter similarity parameter determines correspondence by sending the difference of time in two corresponding batches
Preset weights parameter.
In some possible embodiments, further includes: m preset weights parameter is determined by following formula, and respectively to m
A similarity parameter is weighted processing, to obtain m Weighted Similarity parameter:
Wherein, taThe time is sent in batch for batch transaction message to be measured；SiFor the i-th similarity in m similarity parameter
Parameter；tiTo send the time in the batch corresponding to the i-th batch historical data of the i-th similarity parameter；ωiJoin for m preset weights
Correspond to the i-th preset weights parameter of the i-th similarity parameter in number；XiIt is similar to correspond to i-th in m Weighted Similarity parameter
Spend the i-th Weighted Similarity parameter of parameter；T is to include taAnd each tiThe duration of the first period inside.
In some possible embodiments, further includes: by the default credit information and/or default category of same monitoring object
Property information determines m preset weights parameter.
In some possible embodiments, further includes: extract same monitoring object before given time on send go through
History transaction data, and similarity threshold is determined according to historical trading data, wherein it is sent on historical trading data in historical trading report
Before text.
In some possible embodiments, historical trading data includes: more for corresponding respectively to n third batch number
Transaction data, and each of n third batch number third batch number is correspondingly provided with repeat business risk label, n is big
In 1 positive integer；And method further include: will successively correspond to more of each third batch number in n third batch number
Transaction data is as lot data to be measured, and using the transaction data in historical trading data in addition to lot data to be measured as surplus
Remaining lot data；According to specified message content, the reference similarity between lot data to be measured and remaining lot data is determined
Index, to obtain the reference index similarity for corresponding to each third batch number；According to corresponding to each third batch
Number reference index similarity and repeat business risk label establish ROC curve, thus according to ROC curve determine similarity threshold
Value.
In some possible embodiments, before establishing ROC curve, method further include: removal value is 0 or 1
With reference to index similarity and corresponding repeat business risk label.
In some possible embodiments, the time is sent to have periodically on historical trading data and historical trading message
Corresponding relationship.
In some possible embodiments, further includes: determine batch transaction message to be measured and historical trading message it
Between index similarity before, the batch number of batch transaction message and historical trading message more to be measured；If it exists with to be measured batch
It measures transaction message and compares one or more historical trading messages with same batch number, then directly determine batch transaction report to be measured
There are repeat business risks for text；If it does not exist with the historical trading report of same batch number compared with batch transaction message to be measured
Text then further executes the index similarity determined between batch transaction message to be measured and historical trading message.
In some possible embodiments, further includes: if batch transaction message to be measured is judged, there are repeat business wind
Danger then sends warning information to same monitoring object；The confirmation message that same monitoring object is sent is received, and according to confirmation message
Repetition judges batch transaction message to be measured with the presence or absence of repeat business risk.
A kind of repeat business Risk Monitoring device, comprising: module is obtained, for obtaining by same monitoring object when specified
Engrave the batch transaction message to be measured sent, and before given time on the historical trading message that send；Similarity module, is used for
According to specified message content, the index similarity between batch transaction message to be measured and historical trading message is determined, wherein refer to
Determining message content includes at least two in following: batch number, Transaction Account number and transaction amount；Judgment module, for passing through
Index similarity is compared with default similarity threshold, to judge batch transaction message to be measured with the presence or absence of repeat business wind
Danger.
In some possible embodiments, obtaining module includes: receiving module, for receiving in given time by same
The batch transaction message to be measured sent in monitoring object；Extraction module, for determining for the first period by preset duration and given time,
And extract same monitoring object within the first period on the historical trading message that send.
In some possible embodiments, similarity module includes: similarity measuring and calculating module, for similar using presetting
Degree algorithm determines the similarity vector between batch transaction message to be measured and historical trading message；Similarity score module, is used for
Using default code of points, index similarity is converted by similarity vector.
In some possible embodiments, similarity measuring and calculating module is used for: based on batch transaction message to be measured and history
Transaction message constructs sparse matrix, and in sparse matrix, the value of each nonzero element is determined by transaction amount, each element
Row label and column label determined respectively by batch number and Transaction Account number；Determine the first sparse vector in sparse matrix and m
M similarity parameter between second sparse vector, and similarity vector is determined by m similarity parameter；Wherein, batch to be measured
Transaction message includes: more transaction messages corresponding to first batch number, correspond in sparse matrix the row of first batch number to
Amount/or column vector are as the first sparse vector；Historical trading message includes: more friendships for corresponding respectively to m second lot number
Easy message, the row vector/or column vector that m second lot number is corresponded respectively in sparse matrix as m the second sparse vectors,
M is positive integer.
In some possible embodiments, similarity measuring and calculating module is further used for: by # { (bi- a) ≠ 0 and # { (bi+
A) ratio and/or difference ≠ 0 } determines m phase between the first sparse vector and m the second sparse vectors in sparse matrix
Like degree parameter, wherein i=1,2 ..., m；Wherein, biIndicate i-th of second sparse vectors in m the second sparse vectors, a
Indicate the first sparse vector, # { (bi- a) ≠ 0 indicate the first sparse vector with it is non-in the difference vector of i-th of second sparse vectors
The number of neutral element, # { (b+a) ≠ 0 } indicates the first sparse vector and i-th second sparse vectors and nonzero element in vector
Number.
Specifically: the m between the first sparse vector in sparse matrix and m the second sparse vectors is determined by following equation
A similarity parameter:
In some possible embodiments, similarity score module is used for: determining the maximum phase in m similarity parameter
Like degree parameter as index similarity.
In some possible embodiments, similarity score module is used for: judging the maximum phase in m similarity parameter
Whether reach preset critical like degree parameter；If maximum similarity parameter reaches preset critical, it is determined that preset critical is
Index similarity；If maximum similarity parameter is not up to preset critical, based on m preset weights parameter respectively to m phase
It is weighted processing like degree parameter, to obtain m Weighted Similarity parameter, and determines the maximum in m Weighted Similarity parameter
Weighted Similarity parameter is as index similarity.
In some possible embodiments, specified message content further includes that the time is sent in batch, similarity score module
It is further used for: for each of m similarity parameter similarity parameter, by sending the time in two corresponding batches
Difference and determine corresponding preset weights parameter.
In some possible embodiments, similarity score module is further used for: determining that m are preset by following formula
Weighting parameter, and processing is weighted to m similarity parameter respectively, to obtain m Weighted Similarity parameter:
Wherein, taThe time is sent in batch for batch transaction message to be measured；SiFor the i-th similarity in m similarity parameter
Parameter；tiTo send the time in the batch corresponding to the i-th batch historical data of the i-th similarity parameter；ωiJoin for m preset weights
Correspond to the i-th preset weights parameter of the i-th similarity parameter in number；XiIt is similar to correspond to i-th in m Weighted Similarity parameter
Spend the i-th Weighted Similarity parameter of parameter；T is to include taAnd each tiThe duration of the first period inside.
In some possible embodiments, similarity score module is further used for: by the default of same monitoring object
Credit information and/or preset attribute information determine m preset weights parameter.
In some possible embodiments, further include similarity threshold module, be specifically used for: extracting same monitoring object
The historical trading data sent on before given time, and similarity threshold is determined according to historical trading data, wherein history is handed over
It is sent before historical trading message in easy data.
In some possible embodiments, historical trading data includes: more for corresponding respectively to n third batch number
Transaction data, and each of n third batch number third batch number is correspondingly provided with repeat business risk label, n is big
In 1 positive integer；And similarity threshold module is further used for: will successively correspond to each in n third batch number the
More transaction data of three batch numbers as lot data to be measured, and by historical trading data in addition to lot data to be measured
Transaction data is as remaining lot data；According to specified message content, determine lot data to be measured and remaining lot data it
Between reference index similarity, thus obtain correspond to each third batch number reference index similarity；According to corresponding to
The reference index similarity and repeat business risk label of each third batch number establish ROC curve, thus according to ROC curve
Determine similarity threshold.
In some possible embodiments, before establishing ROC curve, similarity threshold module is further used for: going
Except the reference index similarity and corresponding repeat business risk label that value is 0 or 1.
In some possible embodiments, the time is sent to have periodically on historical trading data and historical trading message
Corresponding relationship.
In some possible embodiments, further include filtering module, be used for: determining batch transaction message to be measured and going through
Before index similarity between history transaction message, the batch number of batch transaction message and historical trading message more to be measured；If
In the presence of compared with batch transaction message to be measured with same batch number one or more historical trading messages, then directly judgement to
Surveying batch transaction message, there are repeat business risks；There is compared with batch transaction message to be measured same batch number if it does not exist
Historical trading message then further executes the index similarity determined between batch transaction message to be measured and historical trading message.
In some possible embodiments, further include warning module, be used for: being handed over if batch transaction message to be measured is judged in the presence of repetition
Easy risk then sends warning information to same monitoring object；The confirmation message that same monitoring object is sent is received, and according to confirmation
Information repeats to judge batch transaction message to be measured with the presence or absence of repeat business risk.
A kind of repeat business risk monitoring system, including such as above-mentioned monitoring device and at least one monitoring object.
A kind of repeat business Risk Monitoring device, comprising: one or more multi-core processor；Memory, for storing
One or more programs；When one or more programs are executed by one or more multi-core processor, so that one or more
Multi-core processor is realized: obtaining the batch transaction message to be measured sent in given time by same monitoring object, and specified
The historical trading message sent on before moment；According to specified message content, batch transaction message to be measured and historical trading are determined
Index similarity between message, wherein specified message content includes at least two in following: batch number, Transaction Account number with
And transaction amount；By being compared to index similarity with default similarity threshold, to judge that batch transaction message to be measured is
It is no that there are repeat business risks.
A kind of computer readable storage medium, computer-readable recording medium storage have program, when program is handled by multicore
When device executes, so that multi-core processor executes such as above-mentioned method.
At least one above-mentioned technical solution that the embodiment of the present application uses can reach following the utility model has the advantages that the present embodiment
In, by calculating same monitoring object in the batch transaction message to be measured sent in given time and in the last period of given time
The index similarity between historical trading message sent on interior, and then by comparing index similarity and preset similarity threshold
Value, more can delicately prompt being monitored the case where part repeat business that may be present in batch transaction message
Repeat business risk, avoids economic loss；Further, during calculating index similarity, the application makes full use of friendship
The easily confidence level of the information promotion index similarity of itself, utilizes the difference and computational short cut between sparse matrix and sparse vector
Similarity calculation process, using send in transaction the time formulate reasonable weighting scheme promoted index similarity calculating it is accurate
Degree；In the formulation process of similarity threshold, the application seeks scheme by formulating reasonable threshold value, is obtained using ROC curve
The higher similarity threshold of confidence level, further ensures the accuracy of repeat business Risk Monitoring.
It should be appreciated that the above description is only an overview of the technical scheme of the present invention, so as to more clearly understand the present invention
Technological means, so as to be implemented in accordance with the contents of the specification.In order to allow above and other objects of the present invention, feature and
Advantage can be more clearly understood, below the special specific embodiment illustrated the present invention.
Detailed description of the invention
By reading the detailed description of following example embodiments, those of ordinary skill in the art are readily apparent that described herein
A little with benefit and other advantage and benefit.Attached drawing is only used for showing the purpose of exemplary embodiment, and is not considered as
Limitation of the present invention.And throughout the drawings, identical component is indicated by the same numeral.In the accompanying drawings:
Fig. 1 is the flow diagram according to the repeat business risk monitoring method of one embodiment of the invention；
Fig. 2 is the flow diagram according to the repeat business risk monitoring method of another embodiment of the present invention；
Fig. 3 is the ROC curve diagram according to the embodiment of the present invention；
Fig. 4 is the structural schematic diagram according to the repeat business Risk Monitoring device of one embodiment of the invention；
Fig. 5 is the structural schematic diagram according to the repeat business Risk Monitoring device of further embodiment of this invention；
Fig. 6 is the schematic diagram according to the computer readable storage medium of one embodiment of the invention.
In the accompanying drawings, identical or corresponding label indicates identical or corresponding part.
Specific embodiment
Exemplary embodiments of the present disclosure are described in more detail below with reference to accompanying drawings.Although showing the disclosure in attached drawing
Exemplary embodiment, it being understood, however, that may be realized in various forms the disclosure without should be by embodiments set forth here
It is limited.On the contrary, these embodiments are provided to facilitate a more thoroughly understanding of the present invention, and can be by the scope of the present disclosure
It is fully disclosed to those skilled in the art.
In the present invention, it should be appreciated that the terms such as " comprising " or " having " are intended to refer to disclosed in this specification
The presence of feature, number, step, behavior, component, part or combinations thereof, and be not intended to other one or more features of exclusion,
Number, step, behavior, component, part or combinations thereof there are a possibility that.
It also should be noted that in the absence of conflict, the feature in embodiment and embodiment in the present invention
It can be combined with each other.The present invention will be described in detail below with reference to the accompanying drawings and embodiments.
The process that Fig. 1 schematically shows the repeat business risk monitoring method 100 of embodiment according to the present invention is shown
It is intended to, it is preferred, but not required, that method shown in FIG. 1 can server, server cluster or backstage trading processing system beyond the clouds
It is executed at system, more specifically, method shown in FIG. 1 can be executed by the specific module being set in Unionpay's system.In the present embodiment,
It is specifically addressed using cloud server as executing subject, it being understood, however, that the application has no specific limit to executing subject
System.
As shown in Figure 1, this method 100 includes:
Step S101, the batch transaction message to be measured sent in given time by same monitoring object is obtained, and in institute
The historical trading message sent on before stating given time；
Wherein, same monitoring object refers to that reality and holder generate trade company or the terminal of transaction, batch transaction to be measured
Message and historical trading message can be the transaction message generated by the multiple types transaction including credit transaction.Same
After a transaction occurs at one monitoring object, not in real time by the transaction message up sending into cloud server, but it is fixed
When in bulk will a period of time in generate more transaction messages packing on send into cloud server.In the present embodiment, together
One monitoring object at multiple time points before given time and given time respectively on send batch transaction message, wherein
The batch transaction message that the monitoring object is sent in given time is appointed as " batch transaction message to be measured ", given time is usual
For nearest moment or current time；By same monitoring object given time for the previous period on the transaction message sent specify
For " historical trading message ", for as the background data in the repeat business risk analysis.
In some possible embodiments, wherein step S101 may further include: receive in given time by same
The batch transaction message to be measured sent in one monitoring object；First period was determined by preset duration and given time, and is extracted same
Monitoring object within the first period on the historical trading message that send.
For example, cloud server is after receiving batch transaction message to be measured, to judge the batch transaction message to be measured
With the presence or absence of repeat business risk, extracted from the database of cloud server the same monitoring object before one day,
Background data of the transaction message for other batches sent in one hour or ten minutes as the repeat business risk analysis.Ying Li
Solution, the transaction message for storing in the database can be batch format or non-batch format, the application are not specifically limited this.
In the application, traded using transaction message of the same monitoring object within a period of time before itself to the batch sent on current
Message is analyzed, and can judge whether current batch transaction message has the wind of repeat business in real time and relatively accurately
Danger.
As shown in Figure 1, this method 100 further include:
Step S102, it according to specified message content, determines between batch transaction message to be measured and historical trading message
Index similarity；
Specifically, specified message content includes at least two in following: batch number, Transaction Account number and transaction amount.
Specifically, for calculating index similarity there are many ways to, for example batch to be measured can be traded based on specified message content
Each transaction message in message and historical trading message is converted to multidimensional characteristic vector, and then is instructed based on historical trading message
Practice and obtain deep learning model, batch transaction message to be measured is input in the deep learning model to export index similarity,
For another example above-mentioned index similarity can be obtained by calculating the modes such as COS distance, Euclidean distance, the application does not make this to have
Body limitation.
In the present embodiment, carry out being not necessarily to other transaction data of additional request when repeat business risk analysis, above-mentioned batch number,
The transaction information such as card number and transaction amount are the message content of transaction itself.Optionally, specified message content may be used also
To include: that information, the application such as time, transaction categories, transaction currency type, tradable commodity type is sent not to limit this specifically in batch
System.
In some possible embodiments, before step S102, method 100 can also include: batch more to be measured
The batch number of transaction message and historical trading message；Wherein, there is compared with batch transaction message to be measured same batch if it exists
Number one or more historical trading messages, then directly determine that there are repeat business risks for batch transaction message to be measured；If not depositing
With the historical trading message of same batch number, then step S102 is further being executed compared with batch transaction message to be measured.
In some possible embodiments, as shown in Fig. 2, wherein step S102 can further comprise:
Step S201, the phase between batch transaction message to be measured and historical trading message is determined using default similarity algorithm
Like degree vector；
In some possible embodiments, wherein step S201 may further include: based on batch to be measured transaction report
Text constructs sparse matrix with historical trading message, and wherein the value of each nonzero element is determined by transaction amount, each member
The row label and column label of element are determined by batch number and Transaction Account number respectively；Determine the first sparse vector and the m in sparse matrix
M similarity parameter between a second sparse vector, and similarity vector is determined by m similarity parameter.
Specifically, batch transaction message to be measured includes: more transaction messages corresponding to first batch number, in sparse matrix
Row vector/or column vector corresponding to first batch number is as the first sparse vector；Historical trading message includes: to correspond respectively to
More transaction messages of m second lot number, correspond respectively in sparse matrix the row vector/of m second lot number or arrange to
For amount as m the second sparse vectors, m is positive integer.
Wherein, the row label of each element determined respectively by batch number with Transaction Account number with column label may is that it is sparse
Each row element in matrix corresponds to same batch number, each column element corresponds to same Transaction Account number；It is also possible to sparse
Each row element in matrix corresponds to same Transaction Account number, each column element corresponds to same batch number.
Wherein, for batch transaction message to be measured, it can be the friendship that same monitoring object is generated according to preset rules
Easily packet, cloud server parses after receiving batch transaction message to be measured obtains more transaction messages.For any batch
Batch Transaction Information for, it will be understood that it is the shared information of more transaction messages, transaction that the time is sent in batch number and batch
Card number and transaction amount are the exclusive information of each transaction message.
It is row label, is specifically described so that Transaction Account number is column label as an example by batch number below:
For example, each transaction message for including by batch transaction message to be measured and historical trading message is row with batch number
Label is arranged by column label of Transaction Account number, to form sparse matrix as follows.Wherein, each row element pair
Same Transaction Account number should be corresponded in same batch number, each column element, be handed over if a certain Transaction Account number exists in a certain batch
The element value of corresponding position is then the transaction amount of the transaction, if a certain Transaction Account number is in a certain batch by easily record
There is no transaction records, then are zero by the element value of corresponding position, are appreciated that by real trade experience in the sparse matrix
In, there may be a certain number of nonzero elements (i.e. real data) for each row and column, and (neutral element does not count a large amount of neutral element
According to without storage).
Specifically, in following sparse matrix, row label NaCorresponding to first batch number, row label N1~NmCorrespond respectively to m
A second lot number, column label C1~CnIt corresponds respectively to included in above-mentioned batch transaction message to be measured and historical trading message
Each transaction message involved in each Transaction Account number, VmnTo correspond to m-th of second batch in historical trading message
Secondary number and Transaction Account number CnTransaction amount, VanTo correspond to Transaction Account number C in batch transaction message to be measurednTrade gold
Volume, and and so on.
In the sparse matrix, row vector corresponding to the first sparse vector a namely first batch a are as follows:
A=(Va1 Va2 … Van)
In the sparse matrix, m the second sparse vector bi, i=1,2 ..., m namely m second lot institute difference
Corresponding m row vector are as follows:
bi=(Vi1 Vi2 … Vin), wherein i=1,2 ..., m
Further, the first sparse vector a and m the second sparse vector b is calculated separatelyiBetween m similarity parameter
Si, wherein i=1,2 ..., m, to obtain similarity vector (S1, S2..., Sm).It, will be to by establishing above-mentioned sparse matrix
The similarity calculating process surveyed between batch transaction message and historical trading message is reduced to more simple similarity between vectors
The process of calculating.
In some possible embodiments, it is possible to further by # { (bi- a) ≠ 0 and # { (bi+ a) ≠ 0 ratio
And/or difference determines that m similarity between the first sparse vector in the sparse matrix and m the second sparse vectors is joined
Number, wherein i=1,2 ..., m.
For example, can be determined by following equation (1) the first sparse vector in sparse matrix and m the second sparse vectors it
Between m similarity parameter:
In above-mentioned formula (1), bjIndicate i-th of second sparse vectors in m the second sparse vectors；A indicates that first is dilute
Dredge vector, # { (bi- a) ≠ 0 indicate of nonzero element in the difference vector of the first sparse vector and i-th of second sparse vectors
Number, # { (bi+ a) ≠ 0 indicate the first sparse vector and i-th second sparse vectors and in vector nonzero element number；Si
Indicate the i-th similarity parameter of i-th of second sparse vectors Yu the first sparse vector；M is positive integer, indicates that m a second is sparse
The quantity of vector and m similarity parameter.
With a=(Va1 Va2 … Van) and b1=(V11 V12 … V1n) for further specifically described.
(b1- a)=(V11-Va1 V12-Va2 … V1n-Van)；
(b1+ a)=(V11+Va1 V12+Va2 … V1n+Van)；
It is appreciated that with Transaction Account number CnFor, if the Transaction Account number is in N1With NaThere are repeat business in two batch numbers
If, then V1n-Van=0, and V1n+Van≠0.In other words, the number of this repeat business, which is counted, enters # { (b1+a)≠
0 }, but statistics does not enter # { (bi-a)≠0}.It is possible to further speculate to obtain, SiValue it is bigger, corresponding two batches
The multiplicity risk of transaction is higher.
It can thus be seen that above-mentioned formula (1) is based on simply repeat business risk identification susceptibility with higher
Statistics, which calculates, to have preferable effect, above-mentioned similarity parameter S to the identification of repeat businessiValue between [0,1], when
When two batches transaction is identical, similarity parameter is 1, and when two batches transaction is entirely different, similarity parameter is 0.
Optionally, the present invention can also determine the first sparse vector a and m the second sparse vector b by other meansiIt
Between m similarity parameter, such as can by calculating Euclidean distance, the calculations such as COS distance determine, the application couple
This is not especially limited.
As shown in Fig. 2, step S102 can further comprise after step S201:
Step S202: using default code of points, index similarity is converted by similarity vector.
In some possible embodiments, the default code of points wherein in step S202 may include: determining m phase
Like the maximum similarity parameter in degree parameter as index similarity.
In some possible embodiments, the default code of points wherein in step S202 can also include: to judge m
Whether the maximum similarity parameter in similarity parameter reaches preset critical；If maximum similarity parameter reaches preset threshold
Value, it is determined that preset critical is index similarity；It is pre- based on m if maximum similarity parameter is not up to preset critical
If weighting parameter is weighted processing to m similarity parameter respectively, to obtain m Weighted Similarity parameter, and determine that m add
The maximum weighted similarity parameter in similarity parameter is weighed as index similarity.
For example, the similarity parameter S according to acquired in above-mentioned formula (1)iValue can be made between [0,1] by 1
Further if maximum similarity parameter reaches 1, illustrate to trade there are two batches identical, usually may be used for preset critical
It is repeated with being deemed to correspond to the two batches transaction of the maximum similarity parameter.If maximum similarity parameter less than 1, is needed into one
Step combines preset weights parameter to be judged that the preset weights parameter can be by sending the factors such as time to determine in batch.
In some possible embodiments, specified message content further includes that the time is sent in batch, wherein step S202 into
One step includes: for each of m similarity parameter similarity parameter, by the difference for sending the time in two corresponding batches
It is worth and determines corresponding preset weights parameter.Since the transaction of time interval lesser two batches is there are duplicate probability is higher, this reality
It applies in example and determines preset weights parameter by using the time difference is sent in the batch traded by two batches, can determine has more Gao Zhun
The index similarity of exactness.
For example, can determine m preset weights parameter (ω by formula (2)1, ω2..., ωm), and it is pre- according to m respectively
If weighting parameter (ω1, ω2..., ωm) respectively to m similarity parameter (S1, S2..., Sm) it is weighted processing, to obtain m
A Weighted Similarity parameter (X1, X2..., Xm)。
Wherein, formula (2) are as follows:
In above-mentioned formula (2), taThe time is sent in batch for batch transaction message to be measured；SiFor in m similarity parameter
The i-th similarity parameter；tiTo send the time in the batch corresponding to the i-th batch historical data of the i-th similarity parameter；ωiFor m
Correspond to the i-th preset weights parameter of the i-th similarity parameter in a preset weights parameter；XiFor in m Weighted Similarity parameter
The i-th Weighted Similarity parameter corresponding to the i-th similarity parameter；T is the duration of the first period；M is positive integer, indicates m phase
Like the number of degree parameter.
It in some possible embodiments, can also include by the default credit information of same monitoring object and/or pre-
If attribute information determines above-mentioned m preset weights parameter.Optionally, the default credit information of same monitoring object is e.g. same
The bank reference of monitoring object scores.
As shown in Figure 1, method 100 further include:
Step S103, index similarity is compared with default similarity threshold, to judge batch transaction message to be measured
With the presence or absence of repeat business risk.
Specifically, above-mentioned repeat business risk is used to indicate in batch transaction message to be measured that there are one or more and history
The transaction message for mutually duplicate repeat business of trading.
For example, for that by the batch transaction message to be measured sent in same monitoring object, can obtain corresponding each time
Index similarity, can be by index similarity compared with default similarity threshold carries out size, if index similarity is more than default
Similarity threshold, then judging batch transaction message to be measured, there are repeat business risks, may further take related Forewarning Measures,
If index similarity is less than default similarity threshold, judge batch transaction message to be measured for arm's length dealing.
In some possible embodiments, method 100 further include: extract same monitoring object before given time on
The historical trading data sent, and similarity threshold is determined according to historical trading data, wherein sent on historical trading data in going through
Before history transaction message.In the present embodiment, the historical trading data based on same monitoring object and obtain similarity threshold tool
There are higher adaptivity and reliability.Optionally, the present embodiment can also obtain similarity by empirical value and experiment value
Threshold value.
In some possible embodiments, the time is sent to have periodically on historical trading data and historical trading message
Corresponding relationship.For example, historical trading data can be by same monitoring object with historical trading message in adjacent two weeks or phase
It is sent in the same period in adjacent two days.
In some possible embodiments, historical trading data includes: more for corresponding respectively to n third batch number
Transaction data, and each of n third batch number third batch number is correspondingly provided with repeat business risk label, n is big
In 1 positive integer.
Further, determine that similarity threshold can specifically include according to historical trading data:
(1) more transaction data of each third batch number in n third batch number will successively be corresponded to as to be measured
Lot data, and using the transaction data in historical trading data in addition to lot data to be measured as remaining lot data；
(2) according to specified message content, the reference similarity between lot data to be measured and remaining lot data is determined
Index, to obtain the reference index similarity for corresponding to each third batch number；
(3) it is established according to the reference index similarity and repeat business risk label that correspond to each third batch number
ROC curve, to determine similarity threshold according to ROC curve；
For example, historical trading data packet can divide the R corresponding to five third batch numbers1~R5, wherein choose R1Make
For lot data to be measured, by remaining R2~R5As remaining lot data, and lot data to be measured and remaining batch is calculated
Index similarity between data, which is used as, refers to index similarity, namely corresponds to R1Reference index similarity, it is specific to calculate
Process is consistent or similar with the step of above calculating index similarity between batch transaction message to be measured and historical trading message,
Details are not described herein by the application.And so on, it can calculate and correspond respectively to R1~R5The reference similarity of five batches refers to
Number.
The ROC curve of establishing in step (3) is described in detail below in conjunction with table 1.
Table 1:
In above table, R1~R5Respectively indicate each of above-mentioned multiple batches of transaction data third batch number, wherein
R1Corresponding repeat business risk label is 0 (namely non-duplicate transaction), and corresponding reference index similarity is 0.3；R3Institute
Corresponding repeat business risk label is 1 (namely repeat business), and corresponding reference index similarity is 0.9, and successively class
It pushes away；Respectively to correspond to R1~R5Reference index similarity as preset threshold carry out precision ratio and recall ratio judgement, determine
For TP, FP, TN, tetra- kinds of situations of FN, wherein if with reference to index similarity >=threshold value, and repeat business risk label=1, determine
For TP；If with reference to index similarity >=threshold value, and repeat business risk label=0, it is determined as FP；If with reference to index similarity <
Threshold value, and repeat business risk label=1, are determined as FN；If being less than threshold value, and repeat business risk mark with reference to index similarity
Label=0, are determined as TN；Further calculate the real rate TPR and false positive rate FPR of each threshold value, wherein TPR=TP/ (TP+
FN), FPR=FP/ (FP+TN).Further, referring to Fig. 3, using FPR as horizontal axis, TPR is the longitudinal axis, according to corresponding to each threshold value
Real rate TPR and false positive rate FPR, obtains ROC curve, and choose point (0,1) corresponding threshold value 0.7 of the curve near the upper left corner
As similarity threshold.
In some possible embodiments, before establishing ROC curve, can also include: removal value be 0 or 1
With reference to index similarity and corresponding repeat business risk label.So as to avoid threshold value from choosing deviation.
In some possible embodiments, if method 100 can also include: that batch transaction message to be measured is judged presence
Repeat business risk then sends warning information to same monitoring object；The confirmation message sent by same monitoring object is received, and
It is repeated to judge batch transaction message to be measured with the presence or absence of repeat business risk according to confirmation message.For example, when index similarity is big
When similarity threshold, Xiang Tongyi monitoring object feeds back warning information, if index similarity reaches preset critical, feeds back more
Strong early warning avoids economic loss so that the risk for reminding same monitoring object that may have repeat business occurs.
In the present embodiment, by calculating batch transaction message to be measured that same monitoring object is sent in given time and referring to
Timing quarters for the previous period on index similarity between the historical trading message that send, and then index similarity by comparing
With preset similarity threshold, can to being monitored the case where part repeat business that may be present in batch transaction message,
Repeat business risk more can be delicately prompted, economic loss is avoided；Further, during calculating index similarity,
The application makes full use of the confidence level of the information promotion index similarity of transaction itself, using between sparse matrix and sparse vector
Difference and computational short cut similarity calculation process, promote similarity using sending the time to formulate reasonable weighting scheme in transaction
The accuracy in computation of index；In the formulation process of similarity threshold, the application seeks scheme by formulating reasonable threshold value, benefit
The higher similarity threshold of confidence level is obtained with ROC curve, further ensures the accuracy of repeat business Risk Monitoring.
Based on the same technical idea, the embodiment of the present invention also provides a kind of repeat business Risk Monitoring device, for holding
Risk trade monitoring method is repeated provided by any of the above-described embodiment of row.Fig. 4 is a kind of repetition provided in an embodiment of the present invention
Transaction risk monitoring device structural schematic diagram.
As shown in figure 4, repeat business Risk Monitoring device 40 includes:
Module 401 is obtained, for obtaining the batch transaction message to be measured sent in given time by same monitoring object, with
And before given time on the historical trading message that send；
Similarity module 402, for determining batch transaction message to be measured and historical trading report according to specified message content
Index similarity between text, wherein specified message content includes at least two in following: batch number, Transaction Account number and
Transaction amount；
Judgment module 403, for by being compared to index similarity with default similarity threshold, to judge to be measured batch
Measuring transaction message whether there is repeat business risk.
In some possible embodiments, obtain module 401 include: receiving module, for given time receive by
The batch transaction message to be measured sent in same monitoring object；Extraction module, for determining first by preset duration and given time
Period, and extract same monitoring object within the first period on the historical trading message that send.
In some possible embodiments, similarity module 402 includes: similarity measuring and calculating module, default for utilizing
Similarity algorithm determines the similarity vector between batch transaction message to be measured and historical trading message；Similarity score module,
For converting index similarity for similarity vector using default code of points.
In some possible embodiments, similarity measuring and calculating module is used for: based on batch transaction message to be measured and history
Transaction message constructs sparse matrix, and in sparse matrix, the value of each nonzero element is determined by transaction amount, each element
Row label and column label determined respectively by batch number and Transaction Account number；Determine the first sparse vector in sparse matrix and m
M similarity parameter between second sparse vector, and similarity vector is determined by m similarity parameter；Wherein, batch to be measured
Transaction message includes: more transaction messages corresponding to first batch number, correspond in sparse matrix the row of first batch number to
Amount/or column vector are as the first sparse vector；Historical trading message includes: more friendships for corresponding respectively to m second lot number
Easy message, the row vector/or column vector that m second lot number is corresponded respectively in sparse matrix as m the second sparse vectors,
M is positive integer.
In some possible embodiments, similarity measuring and calculating module is further used for: by # { (bi- a) ≠ 0 and # { (bi+
A) ratio and/or difference ≠ 0 } determines m phase between the first sparse vector and m the second sparse vectors in sparse matrix
Like degree parameter, wherein i=1,2 ..., m；Wherein, biIndicate i-th of second sparse vectors in m the second sparse vectors, a
Indicate the first sparse vector, # { (bi- a) ≠ 0 indicate the first sparse vector with it is non-in the difference vector of i-th of second sparse vectors
The number of neutral element, # { (b+a) ≠ 0 } indicates the first sparse vector and i-th second sparse vectors and nonzero element in vector
Number.
Specifically: the m between the first sparse vector in sparse matrix and m the second sparse vectors is determined by following equation
A similarity parameter:
In some possible embodiments, similarity score module is used for: determining the maximum phase in m similarity parameter
Like degree parameter as index similarity.
In some possible embodiments, similarity score module is used for: judging the maximum phase in m similarity parameter
Whether reach preset critical like degree parameter；If maximum similarity parameter reaches preset critical, it is determined that preset critical is
Index similarity；If maximum similarity parameter is not up to preset critical, based on m preset weights parameter respectively to m phase
It is weighted processing like degree parameter, to obtain m Weighted Similarity parameter, and determines the maximum in m Weighted Similarity parameter
Weighted Similarity parameter is as index similarity.
In some possible embodiments, specified message content further includes that the time is sent in batch, similarity score module
It is further used for: for each of m similarity parameter similarity parameter, by sending the time in two corresponding batches
Difference and determine corresponding preset weights parameter.
In some possible embodiments, similarity score module is further used for: determining that m are preset by following formula
Weighting parameter, and processing is weighted to m similarity parameter respectively, to obtain m Weighted Similarity parameter:
Wherein, taThe time is sent in batch for batch transaction message to be measured；SiFor the i-th similarity in m similarity parameter
Parameter；tiTo send the time in the batch corresponding to the i-th batch historical data of the i-th similarity parameter；ωiJoin for m preset weights
Correspond to the i-th preset weights parameter of the i-th similarity parameter in number；XiIt is similar to correspond to i-th in m Weighted Similarity parameter
Spend the i-th Weighted Similarity parameter of parameter；T is to include taAnd each tiThe duration of the first period inside.
In some possible embodiments, similarity score module is further used for: by the default of same monitoring object
Credit information and/or preset attribute information determine m preset weights parameter.
In some possible embodiments, device 40 further includes similarity threshold module, is specifically used for: extracting same prison
Survey object before given time on the historical trading data that send, and similarity threshold is determined according to historical trading data, wherein
It is sent on historical trading data before historical trading message.
In some possible embodiments, historical trading data includes: more for corresponding respectively to n third batch number
Transaction data, and each of n third batch number third batch number is correspondingly provided with repeat business risk label, n is big
In 1 positive integer；And similarity threshold module is further used for: will successively correspond to each in n third batch number the
More transaction data of three batch numbers as lot data to be measured, and by historical trading data in addition to lot data to be measured
Transaction data is as remaining lot data；According to specified message content, determine lot data to be measured and remaining lot data it
Between reference index similarity, thus obtain correspond to each third batch number reference index similarity；According to corresponding to
The reference index similarity and repeat business risk label of each third batch number establish ROC curve, thus according to ROC curve
Determine similarity threshold.
In some possible embodiments, before establishing ROC curve, similarity threshold module is further used for: going
Except the reference index similarity and corresponding repeat business risk label that value is 0 or 1.
In some possible embodiments, the time is sent to have periodically on historical trading data and historical trading message
Corresponding relationship.
In some possible embodiments, device 40 further includes filtering module, for determining batch transaction report to be measured
Before index similarity between text and historical trading message, the batch of batch transaction message and historical trading message more to be measured
Number；If it exists with one or more historical trading messages of same batch number compared with batch transaction message to be measured, then directly
Determine that there are repeat business risks for batch transaction message to be measured；Have compared with batch transaction message to be measured with a batch if it does not exist
Secondary number historical trading message is then further determined between batch transaction message to be measured and historical trading message by similarity module
Index similarity.In some possible embodiments, device 40 further includes warning module, is used for: if batch to be measured is traded
Message is judged that there are repeat business risks, then sends warning information to same monitoring object；Same monitoring object is received to send
Confirmation message, and repeated to judge batch transaction message to be measured with the presence or absence of repeat business risk according to confirmation message.
In the present embodiment, by calculating batch transaction message to be measured that same monitoring object is sent in given time and referring to
Timing quarters for the previous period on index similarity between the historical trading message that send, and then index similarity by comparing
With preset similarity threshold, can to being monitored the case where part repeat business that may be present in batch transaction message,
Repeat business risk more can be delicately prompted, economic loss is avoided；Further, during calculating index similarity,
The application makes full use of the confidence level of the information promotion index similarity of transaction itself, using between sparse matrix and sparse vector
Difference and computational short cut similarity calculation process, promote similarity using sending the time to formulate reasonable weighting scheme in transaction
The accuracy in computation of index；In the formulation process of similarity threshold, the application seeks scheme by formulating reasonable threshold value, benefit
The higher similarity threshold of confidence level is obtained with ROC curve, further ensures the accuracy of repeat business Risk Monitoring.
Based on the same technical idea, the embodiment of the present invention also provides a kind of repeat business risk monitoring system, including such as
The upper monitoring device and at least one monitoring object.
Person of ordinary skill in the field it is understood that various aspects of the invention can be implemented as equipment, method or
Computer readable storage medium.Therefore, various aspects of the invention can be embodied in the following forms, it may be assumed that complete hardware
The embodiment party combined in terms of embodiment, complete Software Implementation (including firmware, microcode etc.) or hardware and software
Formula may be collectively referred to as " circuit ", " module " or " equipment " here.
In some possible embodiments, a kind of repeat business Risk Monitoring device of the invention can include at least one
A or multiple processors and at least one processor.Wherein, the memory is stored with program, when described program is described
When processor executes, so that the processor executes step as shown in Figure 1:
Step S101: the batch transaction message to be measured sent in given time by same monitoring object is obtained, and is being referred to
The historical trading message sent on before timing quarter；
Step S102: it according to specified message content, determines between batch transaction message to be measured and historical trading message
Index similarity, wherein specified message content includes at least two in following: batch number, Transaction Account number and transaction amount；
Step S103: by being compared to index similarity with default similarity threshold, to judge batch transaction to be measured
Message whether there is repeat business risk.
The repeat business Risk Monitoring device 5 of this embodiment according to the present invention is described referring to Fig. 5.Fig. 5
The device 5 of display is only an example, should not function to the embodiment of the present invention and use scope bring any restrictions.
As shown in figure 5, device 5 can be showed in the form of universal computing device, including but not limited to: at least one processing
Device 10, at least one processor 20, the bus 60 for connecting distinct device component.
Bus 60 includes data/address bus, address bus and control bus.
Memory 20 may include volatile memory, such as random access memory (RAM) 21 and/or cache are deposited
Reservoir 22 can further include read-only memory (ROM) 23.
Memory 20 can also include program module 24, and such program module 24 includes but is not limited to: operation equipment, one
It can in a or multiple application programs, other program modules and program data, each of these examples or certain combination
It can include the realization of network environment.
Device 5 can also be communicated with one or more external equipments 2 (such as keyboard, sensing equipment, bluetooth equipment etc.),
It can be communicated with one or more other equipment.This communication can be carried out by input/output (I/O) interface 40, and
It is shown on display unit 30.Also, device 5 can also pass through network adapter 50 and one or more network (example
Such as local area network (LAN), wide area network (WAN) and/or public network, such as internet) communication.As shown, network adapter 50
It is communicated by bus 60 with other modules in device 5.It should be understood that although not shown in the drawings, but can be used with coupling apparatus 5
Other hardware and/or software module, including but not limited to: microcode, device driver, redundant processing unit, external disk drive
Dynamic array, RAID device, tape drive and data backup storage equipment etc..
Fig. 6 shows a kind of computer readable storage medium, for executing method as described above.
In some possible embodiments, various aspects of the invention are also implemented as a kind of computer-readable storage
The form of medium comprising program code, when said program code is when being executed by processor, said program code is for making institute
It states processor and executes method described above.
Method described above include shown in drawings above with unshowned multiple operations and step, here will not
It repeats again.
The computer readable storage medium can be using any combination of one or more readable mediums.Readable medium can
To be readable signal medium or readable storage medium storing program for executing.Readable storage medium storing program for executing for example may be-but not limited to-electricity, magnetic,
Optical, electromagnetic, the equipment of infrared ray or semiconductor, equipment or device, or any above combination.Readable storage medium storing program for executing is more
Specific example (non exhaustive list) includes: the electrical connection with one or more conducting wires, portable disc, hard disk, deposits at random
It is access to memory (RAM), read-only memory (ROM), erasable programmable read only memory (EPROM or flash memory), optical fiber, portable
Compact disk read-only memory (CD-ROM), light storage device, magnetic memory device or above-mentioned any appropriate combination.
As shown in fig. 6, describing the computer readable storage medium 6 of embodiment according to the present invention, can use
Portable compact disc read only memory (CD-ROM) and including program code, and can be on terminal device, such as PC
Operation.However, computer readable storage medium of the invention is without being limited thereto, in this document, readable storage medium storing program for executing, which can be, appoints
What include or the tangible medium of storage program that the program can be commanded and execute equipment, equipment or device use or and its
It is used in combination.
The program for executing operation of the present invention can be write with any combination of one or more programming languages
Code, described program design language include object oriented program language-Java, C++ etc., further include conventional
Procedural programming language-such as " C " language or similar programming language.Program code can be fully in user
It is executed in calculating equipment, partly execution part executes on a remote computing or completely long-range on a user device
It calculates and is executed on equipment or server.In the situation for being related to remote computing device, remote computing device can be by any number of
The network of class --- it is connected to user calculating equipment including local area network (LAN) or wide area network (WAN)-, or, it may be connected to
External computing device (such as being connected using ISP by internet).
In addition, although describing the operation of the method for the present invention in the accompanying drawings with particular order, this do not require that or
Hint must execute these operations in this particular order, or have to carry out shown in whole operation be just able to achieve it is desired
As a result.Additionally or alternatively, it is convenient to omit multiple steps are merged into a step and executed by certain steps, and/or by one
Step is decomposed into execution of multiple steps.
Although detailed description of the preferred embodimentsthe spirit and principles of the present invention are described by reference to several, it should be appreciated that, this
It is not limited to the specific embodiments disclosed for invention, does not also mean that the feature in these aspects cannot to the division of various aspects
Combination is benefited to carry out, this to divide the convenience merely to statement.The present invention is directed to cover appended claims spirit and
Included various modifications and equivalent arrangements in range.
Claims (35)
1. a kind of repeat business risk monitoring method characterized by comprising
The batch transaction message to be measured sent in given time by same monitoring object is obtained, and before the given time
On the historical trading message that send；
According to specified message content, the similarity between the batch transaction message to be measured and the historical trading message is determined
Index, wherein the specified message content includes at least two in following: batch number, Transaction Account number and transaction amount；
By being compared to the index similarity with default similarity threshold, to judge that the batch transaction message to be measured is
It is no that there are repeat business risks.
2. by method described in claim 1, which is characterized in that the acquisition was sent in given time by same monitoring object
Batch transaction message to be measured, and before the given time on the historical trading message that send include:
The batch transaction message to be measured by sending in the same monitoring object is received in the given time；
First period was determined by preset duration and the given time, and extracts the same monitoring object in first period
The historical trading message sent on interior.
3. by method described in claim 1, which is characterized in that determine the batch transaction message to be measured and the historical trading
Index similarity between message includes:
The similarity between the batch transaction message to be measured and the historical trading message is determined using default similarity algorithm
Vector；
Using default code of points, the index similarity is converted by the similarity vector.
4. by method as claimed in claim 3, which is characterized in that described to determine the batch to be measured using default similarity algorithm
Similarity vector between transaction message and the historical trading message includes:
Sparse matrix is constructed based on the batch transaction message to be measured and the historical trading message, in the sparse matrix, often
The value of one nonzero element is determined that the row label and column label of each element are respectively by the batch by the transaction amount
Number with the Transaction Account number determine；
Determine m similarity parameter between the first sparse vector in the sparse matrix and m the second sparse vectors, and by
The m similarity parameter determines the similarity vector；
Wherein, the batch transaction message to be measured includes: more transaction messages corresponding to first batch number, the sparse matrix
In correspond to the first batch number row vector/or column vector as first sparse vector；The historical trading message
Include: the more transaction messages for corresponding respectively to m second lot number, the m the is corresponded respectively in the sparse matrix
For the row vector/or column vector of two batch numbers as the m the second sparse vectors, the m is positive integer.
5. by method as claimed in claim 4, which is characterized in that further include:
By # { (bi- a) ≠ 0 and # { (bi+ a) ≠ 0 ratio and/or difference determine in the sparse matrix first it is sparse to
M similarity parameter between amount and m the second sparse vectors, wherein i=1,2 ..., m；
Wherein, the biIndicate i-th of second sparse vectors in the m the second sparse vectors, a indicates described first
Sparse vector, the # { (bi- a) ≠ 0 indicate the difference vector of first sparse vector and i-th of second sparse vectors
The number of middle nonzero element, the # { (b+a) ≠ 0 } indicate first sparse vector and i-th of second sparse vectors
With the number of nonzero element in vector.
6. by method as claimed in claim 4, which is characterized in that the default code of points includes:
Determine the maximum similarity parameter in the m similarity parameter as the index similarity.
7. by method as claimed in claim 4, which is characterized in that the default code of points further include:
Judge whether the maximum similarity parameter in the m similarity parameter reaches preset critical；
If the maximum similarity parameter reaches the preset critical, it is determined that the preset critical is that the similarity refers to
Number；
If the maximum similarity parameter is not up to the preset critical, based on m preset weights parameter respectively to the m
A similarity parameter is weighted processing, to obtain m Weighted Similarity parameter, and determines the m Weighted Similarity parameter
In maximum weighted similarity parameter as the index similarity.
8. by method of claim 7, which is characterized in that the specified message content further includes sending the time in batch, institute
State method further include:
For each of m similarity parameter similarity parameter, by the difference for sending the time in two corresponding batches
It is worth and determines the corresponding preset weights parameter.
9. by method according to any one of claims 8, which is characterized in that further include:
The m preset weights parameter is determined by following formula, and processing is weighted to the m similarity parameter respectively,
To obtain the m Weighted Similarity parameter:
Wherein, taThe time is sent in batch for the batch transaction message to be measured；SiFor the i-th phase in the m similarity parameter
Like degree parameter；tiTo send the time in the batch corresponding to the i-th batch historical data of the i-th similarity parameter；ωiFor the m
Correspond to the i-th preset weights parameter of the i-th similarity parameter in a preset weights parameter；XiIt is similar for described m weighting
It spends in parameter and corresponds to the i-th Weighted Similarity parameter of the i-th similarity parameter；T is to include the taAnd each institute
State tiThe duration of the first period inside.
10. by method of claim 7, which is characterized in that further include: believed by the default credit of the same monitoring object
Breath and/or preset attribute information determine the m preset weights parameter.
11. by method of any of claims 1-10, which is characterized in that further include:
Extract the same monitoring object before the given time on the historical trading data that send, and handed over according to the history
Easy data determine the similarity threshold, wherein are sent before the historical trading message on the historical trading data.
12. the method as described in claim 11, which is characterized in that the historical trading data includes: to correspond respectively to n the
More transaction data of three batch numbers, and each of described n third batch number third batch number is correspondingly provided with repetition
Transaction risk label, the n are the positive integer greater than 1；And
The method also includes:
More transaction data of each third batch number in the n third batch number will successively be corresponded to as batch to be measured
Data, and using the transaction data in the historical trading data in addition to the lot data to be measured as remaining lot data；
According to the specified message content, the reference phase between the lot data to be measured and the remaining lot data is determined
Like degree index, to obtain the reference index similarity for corresponding to each third batch number；
According to the reference index similarity and the repeat business risk label for corresponding to each third batch number
ROC curve is established, to determine the similarity threshold according to the ROC curve.
13. the method as described in claim 12, which is characterized in that before establishing the ROC curve, the method is also wrapped
It includes:
Remove reference index similarity and corresponding repeat business risk label that value is 0 or 1.
14. the method as described in claim 11, which is characterized in that the historical trading data and the historical trading message
On send the time that there is periodical corresponding relationship.
15. by method described in claim 1, which is characterized in that further include:
Before index similarity between the determination batch transaction message to be measured and the historical trading message, compare
The batch number of the batch transaction message to be measured and the historical trading message；
If it exists with one or more historical trading messages of same batch number compared with the batch transaction message to be measured, then
Directly determine that there are repeat business risks for the batch transaction message to be measured；
It is then further held compared with the batch transaction message to be measured with the historical trading message of same batch number if it does not exist
Index similarity between the row determination batch transaction message to be measured and the historical trading message.
16. the method as described in claim 1 or 15, which is characterized in that further include:
If the batch transaction message to be measured is judged there are repeat business risk, early warning is sent to the same monitoring object
Information；
The confirmation message that the same monitoring object is sent is received, and is repeated to judge the batch to be measured according to the confirmation message
Transaction message whether there is repeat business risk.
17. a kind of repeat business Risk Monitoring device characterized by comprising
Module is obtained, for obtaining the batch transaction message to be measured sent in given time by same monitoring object, and in institute
The historical trading message sent on before stating given time；
Similarity module, for determining the batch transaction message to be measured and the historical trading according to specified message content
Index similarity between message, wherein the specified message content includes at least two in following: batch number, transaction account
Number and transaction amount；
Judgment module, it is described to be measured to judge for by being compared to the index similarity with default similarity threshold
Batch transaction message whether there is repeat business risk.
18. the device as described in claim 17, which is characterized in that the acquisition module includes:
Receiving module, for receiving the batch transaction message to be measured by sending in the same monitoring object in the given time；
Extraction module for determining for the first period by preset duration and the given time, and extracts the same monitoring object
The historical trading message sent within first period.
19. the device as described in claim 17, which is characterized in that the similarity module includes:
Similarity calculates module, for determining that the batch transaction message to be measured and the history are handed over using default similarity algorithm
Similarity vector between easy message；
Similarity score module, for converting the index similarity for the similarity vector using default code of points.
20. the device as described in claim 19, which is characterized in that the similarity measuring and calculating module is used for:
Sparse matrix is constructed based on the batch transaction message to be measured and the historical trading message, in the sparse matrix, often
The value of one nonzero element is determined that the row label and column label of each element are respectively by the batch by the transaction amount
Number with the Transaction Account number determine；
Determine m similarity parameter between the first sparse vector in the sparse matrix and m the second sparse vectors, and by
The m similarity parameter determines the similarity vector；
Wherein, the batch transaction message to be measured includes: more transaction messages corresponding to first batch number, the sparse matrix
In correspond to the first batch number row vector/or column vector as first sparse vector；The historical trading message
Include: the more transaction messages for corresponding respectively to m second lot number, the m the is corresponded respectively in the sparse matrix
For the row vector/or column vector of two batch numbers as the m the second sparse vectors, the m is positive integer.
21. the device as described in claim 20, which is characterized in that the similarity measuring and calculating module is further used for:
By # { (bi- a) ≠ 0 and # { (bi+ a) ≠ 0 ratio and/or difference determine in the sparse matrix first it is sparse to
M similarity parameter between amount and m the second sparse vectors, wherein i=1,2 ..., m；
Wherein, the biIndicate i-th of second sparse vectors in the m the second sparse vectors, a indicates described first
Sparse vector, the # { (bi- a) ≠ 0 indicate the difference vector of first sparse vector and i-th of second sparse vectors
The number of middle nonzero element, the # { (b+a) ≠ 0 } indicate first sparse vector and i-th of second sparse vectors
With the number of nonzero element in vector.
22. the device as described in claim 20, which is characterized in that the similarity score module is used for:
Determine the maximum similarity parameter in the m similarity parameter as the index similarity.
23. the device as described in claim 20, which is characterized in that the similarity score module is used for:
Judge whether the maximum similarity parameter in the m similarity parameter reaches preset critical；
If the maximum similarity parameter reaches the preset critical, it is determined that the preset critical is that the similarity refers to
Number；
If the maximum similarity parameter is not up to the preset critical, based on m preset weights parameter respectively to the m
A similarity parameter is weighted processing, to obtain m Weighted Similarity parameter, and determines the m Weighted Similarity parameter
In maximum weighted similarity parameter as the index similarity.
24. the device as described in claim 23, which is characterized in that the specified message content further includes that the time is sent in batch,
The similarity score module is further used for:
For each of m similarity parameter similarity parameter, by the difference for sending the time in two corresponding batches
It is worth and determines the corresponding preset weights parameter.
25. the device as described in claim 24, which is characterized in that the similarity score module is further used for:
The m preset weights parameter is determined by following formula, and processing is weighted to the m similarity parameter respectively,
To obtain the m Weighted Similarity parameter:
Wherein, taThe time is sent in batch for the batch transaction message to be measured；SiFor the i-th phase in the m similarity parameter
Like degree parameter；tiTo send the time in the batch corresponding to the i-th batch historical data of the i-th similarity parameter；ωiFor the m
Correspond to the i-th preset weights parameter of the i-th similarity parameter in a preset weights parameter；XiIt is similar for described m weighting
It spends in parameter and corresponds to the i-th Weighted Similarity parameter of the i-th similarity parameter；T is to include the taAnd each institute
State tiThe duration of the first period inside.
26. the device as described in claim 23, which is characterized in that the similarity score module is further used for: by described
The default credit information and/or preset attribute information of same monitoring object determine the m preset weights parameter.
27. the device as described in any one of claim 17-26, which is characterized in that further include similarity threshold module, specifically
For:
Extract the same monitoring object before the given time on the historical trading data that send, and handed over according to the history
Easy data determine the similarity threshold, wherein are sent before the historical trading message on the historical trading data.
28. the device as described in claim 27, which is characterized in that the historical trading data includes: to correspond respectively to n the
More transaction data of three batch numbers, and each of described n third batch number third batch number is correspondingly provided with repetition
Transaction risk label, the n are the positive integer greater than 1；And
The similarity threshold module is further used for:
More transaction data of each third batch number in the n third batch number will successively be corresponded to as batch to be measured
Data, and using the transaction data in the historical trading data in addition to the lot data to be measured as remaining lot data；
According to the specified message content, the reference phase between the lot data to be measured and the remaining lot data is determined
Like degree index, to obtain the reference index similarity for corresponding to each third batch number；
According to the reference index similarity and the repeat business risk label for corresponding to each third batch number
ROC curve is established, to determine the similarity threshold according to the ROC curve.
29. the device as described in claim 28, which is characterized in that before establishing the ROC curve, the similarity threshold
Module is further used for:
Remove reference index similarity and corresponding repeat business risk label that value is 0 or 1.
30. the device as described in claim 27, which is characterized in that the historical trading data and the historical trading message
On send the time that there is periodical corresponding relationship.
31. the device as described in claim 17, which is characterized in that further include filtering module, be used for:
Before index similarity between the determination batch transaction message to be measured and the historical trading message, compare
The batch number of the batch transaction message to be measured and the historical trading message；
If it exists with one or more historical trading messages of same batch number compared with the batch transaction message to be measured, then
Directly determine that there are repeat business risks for the batch transaction message to be measured；
It is then further held compared with the batch transaction message to be measured with the historical trading message of same batch number if it does not exist
Index similarity between the row determination batch transaction message to be measured and the historical trading message.
32. the device as described in claim 17 or 31, which is characterized in that further include warning module, be used for:
If the batch transaction message to be measured is judged there are repeat business risk, early warning is sent to the same monitoring object
Information；
The confirmation message that the same monitoring object is sent is received, and is repeated to judge the batch to be measured according to the confirmation message
Transaction message whether there is repeat business risk.
33. a kind of repeat business risk monitoring system, which is characterized in that including as described in any one of claim 17-32
Monitoring device and at least one monitoring object.
34. a kind of repeat business Risk Monitoring device characterized by comprising
One or more multi-core processor；
Memory, for storing one or more programs；
When one or more of programs are executed by one or more of multi-core processors, so that one or more of
Multi-core processor is realized:
The batch transaction message to be measured sent in given time by same monitoring object is obtained, and before the given time
On the historical trading message that send；
According to specified message content, the similarity between the batch transaction message to be measured and the historical trading message is determined
Index, wherein the specified message content includes at least two in following: batch number, Transaction Account number and transaction amount；
By being compared to the index similarity with default similarity threshold, to judge that the batch transaction message to be measured is
It is no that there are repeat business risks.
35. a kind of computer readable storage medium, the computer-readable recording medium storage has program, when described program is more
When core processor executes, so that the multi-core processor executes the method as described in any one of claim 1-16.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910405528.XA CN110135856B (en) | 2019-05-16 | 2019-05-16 | Repeated transaction risk monitoring method, device and computer readable storage medium |
| PCT/CN2020/087550 WO2020228530A1 (en) | 2019-05-16 | 2020-04-28 | Repeated transaction risk monitoring method and device, and computer readable storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910405528.XA CN110135856B (en) | 2019-05-16 | 2019-05-16 | Repeated transaction risk monitoring method, device and computer readable storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110135856A true CN110135856A (en) | 2019-08-16 |
| CN110135856B CN110135856B (en) | 2023-10-24 |
Family
ID=67574321
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910405528.XA Active CN110135856B (en) | 2019-05-16 | 2019-05-16 | Repeated transaction risk monitoring method, device and computer readable storage medium |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN110135856B (en) |
| WO (1) | WO2020228530A1 (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110705992A (en) * | 2019-09-27 | 2020-01-17 | 支付宝(杭州)信息技术有限公司 | Similarity evaluation method and device for risk prevention and control strategy |
| CN111144975A (en) * | 2019-12-06 | 2020-05-12 | 港融科技有限公司 | Order matching method, server and computer readable storage medium |
| CN111429277A (en) * | 2020-03-18 | 2020-07-17 | 中国工商银行股份有限公司 | Repeated transaction prediction method and system |
| WO2020228530A1 (en) * | 2019-05-16 | 2020-11-19 | 中国银联股份有限公司 | Repeated transaction risk monitoring method and device, and computer readable storage medium |
| WO2021032056A1 (en) * | 2019-08-21 | 2021-02-25 | 深圳前海微众银行股份有限公司 | Method and apparatus for processing batch tasks, computing device and storage medium |
| CN112465638A (en) * | 2020-11-16 | 2021-03-09 | 中科金审(北京)科技有限公司 | Link tracking method for fund transaction split combination |
| CN113516555A (en) * | 2021-04-26 | 2021-10-19 | 中国工商银行股份有限公司 | Repeated service transaction detection method and device |
| WO2022105183A1 (en) * | 2020-11-20 | 2022-05-27 | 平安科技（深圳）有限公司 | User clustering method, apparatus and device |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113362157B (en) * | 2021-05-27 | 2024-02-09 | 中国银联股份有限公司 | Abnormal node identification method, model training method, device and storage medium |
| CN116295762B (en) * | 2023-02-09 | 2024-05-03 | 山东黄河河务局山东黄河信息中心 | Root stone position monitoring system and method based on wireless sensing technology |
| CN116308215B (en) * | 2023-05-17 | 2023-07-21 | 云账户技术(天津)有限公司 | Method and device for generating batch outgoing information and related equipment |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103049851A (en) * | 2012-12-27 | 2013-04-17 | 中国建设银行股份有限公司 | Transaction data-based anti-fraud monitoring method and device |
| CN106157140A (en) * | 2014-09-25 | 2016-11-23 | 天逸财金科技服务股份有限公司 | transaction document comparison method and system |
| US20170300919A1 (en) * | 2014-12-30 | 2017-10-19 | Alibaba Group Holding Limited | Transaction risk detection method and apparatus |
| CN108122114A (en) * | 2017-12-25 | 2018-06-05 | 同济大学 | For abnormal repeat business fraud detection method, system, medium and equipment |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7177846B2 (en) * | 2002-07-29 | 2007-02-13 | Checkfree Corporation | Technique for account authentication |
| CN108133373A (en) * | 2018-01-04 | 2018-06-08 | 交通银行股份有限公司 | Seek the method and device for the adventure account for relating to machine behavior |
| CN109191136A (en) * | 2018-09-05 | 2019-01-11 | 北京芯盾时代科技有限公司 | A kind of e-bank is counter to cheat method and device |
| CN110135856B (en) * | 2019-05-16 | 2023-10-24 | 中国银联股份有限公司 | Repeated transaction risk monitoring method, device and computer readable storage medium |
-
2019
- 2019-05-16 CN CN201910405528.XA patent/CN110135856B/en active Active
-
2020
- 2020-04-28 WO PCT/CN2020/087550 patent/WO2020228530A1/en active Application Filing
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103049851A (en) * | 2012-12-27 | 2013-04-17 | 中国建设银行股份有限公司 | Transaction data-based anti-fraud monitoring method and device |
| CN106157140A (en) * | 2014-09-25 | 2016-11-23 | 天逸财金科技服务股份有限公司 | transaction document comparison method and system |
| US20170300919A1 (en) * | 2014-12-30 | 2017-10-19 | Alibaba Group Holding Limited | Transaction risk detection method and apparatus |
| CN108122114A (en) * | 2017-12-25 | 2018-06-05 | 同济大学 | For abnormal repeat business fraud detection method, system, medium and equipment |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2020228530A1 (en) * | 2019-05-16 | 2020-11-19 | 中国银联股份有限公司 | Repeated transaction risk monitoring method and device, and computer readable storage medium |
| WO2021032056A1 (en) * | 2019-08-21 | 2021-02-25 | 深圳前海微众银行股份有限公司 | Method and apparatus for processing batch tasks, computing device and storage medium |
| CN110705992A (en) * | 2019-09-27 | 2020-01-17 | 支付宝(杭州)信息技术有限公司 | Similarity evaluation method and device for risk prevention and control strategy |
| CN111144975A (en) * | 2019-12-06 | 2020-05-12 | 港融科技有限公司 | Order matching method, server and computer readable storage medium |
| CN111144975B (en) * | 2019-12-06 | 2023-09-12 | 港融科技有限公司 | Order matching method, server and computer readable storage medium |
| CN111429277A (en) * | 2020-03-18 | 2020-07-17 | 中国工商银行股份有限公司 | Repeated transaction prediction method and system |
| CN111429277B (en) * | 2020-03-18 | 2023-11-24 | 中国工商银行股份有限公司 | Repeat transaction prediction method and system |
| CN112465638A (en) * | 2020-11-16 | 2021-03-09 | 中科金审(北京)科技有限公司 | Link tracking method for fund transaction split combination |
| WO2022105183A1 (en) * | 2020-11-20 | 2022-05-27 | 平安科技（深圳）有限公司 | User clustering method, apparatus and device |
| CN113516555A (en) * | 2021-04-26 | 2021-10-19 | 中国工商银行股份有限公司 | Repeated service transaction detection method and device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110135856B (en) | 2023-10-24 |
| WO2020228530A1 (en) | 2020-11-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110135856A (en) | A kind of repeat business risk monitoring method, device and computer readable storage medium | |
| CN109242261B (en) | Method for evaluating security risk based on big data and terminal equipment | |
| CN103370722B (en) | The system and method that actual volatility is predicted by small echo and nonlinear kinetics | |
| CN111127178A (en) | Data processing method and device, storage medium and electronic equipment | |
| US20180341934A1 (en) | Atm withdrawal query method and apparatus | |
| CN107679734A (en) | It is a kind of to be used for the method and system without label data classification prediction | |
| CN104866408A (en) | Capacity prediction method and device for application system | |
| CN113095408A (en) | Risk determination method and device and server | |
| CN110162458A (en) | Test data building method, device and storage medium | |
| CN108111368A (en) | The function test method and device of a kind of transaction system | |
| JP2023517338A (en) | ABNORMAL ACTION DETECTION METHOD, APPARATUS, ELECTRONIC DEVICE, AND COMPUTER PROGRAM | |
| CN114595765A (en) | Data processing method and device, electronic equipment and storage medium | |
| CN112035729A (en) | Wrong account processing method, device, equipment and medium | |
| US20220327450A1 (en) | Method for increasing or decreasing number of workers and inspectors in crowdsourcing-based project for creating artificial intelligence learning data | |
| CN113849618A (en) | Strategy determination method and device based on knowledge graph, electronic equipment and medium | |
| CN112801688A (en) | Method and device for positioning reason of estimation failure | |
| CN113807391A (en) | Task model training method and device, electronic equipment and storage medium | |
| CN116644372B (en) | Account type determining method and device, electronic equipment and storage medium | |
| CN110866557B (en) | Data evaluation method and device, storage medium and electronic device | |
| US20220358506A1 (en) | Methods and systems for resolving automated teller machine errors | |
| CN116433392A (en) | Reinsurance measuring and calculating method, device and equipment | |
| CN112766552A (en) | Method and device for optimizing internet architecture and electronic equipment | |
| CN115630960A (en) | Information display method and device, electronic equipment and medium | |
| CN117236956A (en) | Method, server and medium for determining service of unmanned network point | |
| CN115689574A (en) | Transaction risk early warning method and device, electronic equipment and storage medium |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |