CN109685647A - The training method of credit fraud detection method and its model, device and server - Google Patents
The training method of credit fraud detection method and its model, device and server Download PDFInfo
- Publication number
- CN109685647A CN109685647A CN201811616968.1A CN201811616968A CN109685647A CN 109685647 A CN109685647 A CN 109685647A CN 201811616968 A CN201811616968 A CN 201811616968A CN 109685647 A CN109685647 A CN 109685647A
- Authority
- CN
- China
- Prior art keywords
- credit
- data
- customer
- node
- fraud detection
- 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
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
- G06Q40/03—Credit; Loans; Processing thereof
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Finance (AREA)
- Accounting & Taxation (AREA)
- General Physics & Mathematics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Development Economics (AREA)
- Economics (AREA)
- Marketing (AREA)
- Strategic Management (AREA)
- Technology Law (AREA)
- General Business, Economics & Management (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The present invention provides a kind of credit fraud detection method and its training method, device and the servers of model, wherein in the training method of credit fraud detection model, determines client's statistical nature vector sum customer relationship map by the credit customer data of acquisition；By the corresponding state tag data of the credit customer data of acquisition, the sequence node in customer relationship map is converted into node label sequence；Then by skip-gram algorithm, node label is Sequence Transformed for the corresponding feature vector of customer relationship map；According to the feature vector of statistical nature vector sum customer relationship map, initial model is trained, until the number of iterations of training meets preset the number of iterations threshold value, obtains credit fraud detection model.Relation map network structure, label characteristics and statistical nature vector is utilized in the credit fraud detection model that the present invention obtains, and can preferably identify the risk of fraud of client, to improve the reliability of risk supervision result.
Description
Technical field
The present invention relates to internet credit air control technical fields, more particularly, to a kind of credit fraud detection method and its mould
Training method, device and the server of type.
Background technique
It is many to interconnect with the development of knowledge mapping and social networks technology and the extensive use in internet area
Net financial institution also starts relation map being applied to internet credit field；For example visualize character relation, with
Indirect labor borrows preceding examination & approval；According to the several derivative variables of expertise summary artificial from map, for example once it is associated with spy
Sign, two degree of linked characters etc., or such as out-degree, in-degree, centrality and PageRank (webpage grade are extracted from social networks
The feature based on graph theory such as not), for applying for scoring or anti-fraud；These methods be all based on the network structure of relation map into
The certain information characteristics of row extract, face to face to the complex network of substantial amounts when, certainly exist during this certain
Loss of learning can not give full expression to the structural information of network, cause only to carry out internet letter by the network structure of relation map
Detection is borrowed, obtained result reliability is poor.
Summary of the invention
In view of this, the purpose of the present invention is to provide the training methods of credit fraud detection method and its model, device
And server, to improve the reliability of credit risk testing result.
In a first aspect, the embodiment of the invention provides a kind of training methods of credit fraud detection model, this method comprises:
Obtain credit customer data and corresponding state tag data；According to credit customer data, client's statistical nature vector sum is determined
Customer relationship map；By the corresponding state tag data of credit customer data, states credit customer data are corresponding client and close
It is that sequence node in map is converted to node label sequence；It is by skip-gram algorithm, node label is Sequence Transformed for visitor
The corresponding feature vector of family relation map；According to the feature vector of statistical nature vector sum customer relationship map, to initial model
It is trained, until the number of iterations of training meets preset the number of iterations threshold value, obtains credit fraud detection model.
With reference to first aspect, the embodiment of the invention provides the first possible embodiments of first aspect, wherein on
Stating customer relationship map includes request for data, attribute data and corresponding customer relationship；Client's statistical nature vector includes row
For statistical nature vector, application statistical nature vector sum relationship statistical nature vector, which passes through following
Mode obtains: obtaining device data, request for data and the social data of credit customer data；By the calculating to device data,
Obtain behavioral statistics feature vector；By the calculating to request for data, application statistical nature vector is obtained；By to social data
Calculating, obtain relationship statistical nature vector.
With reference to first aspect, the embodiment of the invention provides second of possible embodiments of first aspect, wherein
By the corresponding state tag data of credit customer data, by the node sequence in the corresponding customer relationship map of credit customer data
Before the step of column are converted to node label sequence, further includes: calculated by the preset relationship weight of customer relationship and random walk
Method obtains the sequence node of customer relationship map；Sequence node includes application node and attribute node.
With reference to first aspect, the embodiment of the invention provides the third possible embodiments of first aspect, wherein on
It states through the corresponding state tag data of credit customer data, by the node in the corresponding customer relationship map of credit customer data
It includes: by the sequence node in the corresponding customer relationship map of credit customer data that sequence, which is converted to the step of node label sequence,
In application node be converted into the corresponding state tag data of credit customer data；By the corresponding customer relationship of credit customer data
The attribute node in sequence node in map is converted into request for data, attribute data or corresponding in customer relationship map
Customer relationship, to obtain the corresponding node label sequence of sequence node.
With reference to first aspect, the embodiment of the invention provides the 4th kind of possible embodiments of first aspect, wherein on
It states through skip-gram algorithm, the step of feature vector corresponding for customer relationship map that node label is Sequence Transformed, packet
Include: by node label sequence inputting to skip-gram model, output obtains the corresponding feature vector of customer relationship map.
With reference to first aspect, the embodiment of the invention provides the 5th kind of possible embodiments of first aspect, wherein on
The feature vector according to statistical nature vector sum customer relationship map is stated, initial model is trained, until the iteration of training
The step of number meets preset the number of iterations threshold value, obtains credit fraud detection model includes: by statistical nature vector sum visitor
The feature vector of family relation map merges；According to the feature vector after merging, using supervised classification algorithm to initial model
It is trained, when trained the number of iterations meets preset the number of iterations threshold value, obtains credit fraud detection model.
Second aspect, the embodiment of the present invention also provide a kind of credit fraud detection method, and this method is applied to configured with letter
The equipment for borrowing fraud detection model；The credit fraud detection model is the model that the training of first aspect the method obtains；The party
Method includes: to obtain credit customer data, corresponding state tag data and customer relationship map to be predicted；By credit customer number
It is input in credit fraud detection model according to, state tag data and customer relationship map, obtains the corresponding fraud of credit customer
Risk.
The third aspect, the embodiment of the present invention also provide a kind of training device of credit fraud detection model, which includes:
Customer data obtains module, for obtaining credit customer data and corresponding state tag data；Map construction module is used for root
It is believed that borrowing customer data, client's statistical nature vector sum customer relationship map is determined；Sequence conversion module, for passing through credit visitor
Sequence node in the corresponding customer relationship map of credit customer data is converted to section by the corresponding state tag data of user data
Point sequence label；Feature vector module, being used for will be Sequence Transformed for client pass by node label by skip-gram algorithm
It is the corresponding feature vector of map；Model training module, for according to the feature of statistical nature vector sum customer relationship map to
Amount, is trained initial model, until trained the number of iterations meets preset the number of iterations threshold value, obtains credit fraud inspection
Survey model.
Fourth aspect, the embodiment of the present invention also provide a kind of credit fraud detection device, which is applied to configured with letter
The equipment for borrowing fraud detection model；The credit fraud detection model is the mould that the training of claim first aspect the method obtains
Type；The device includes: data acquisition module, for obtain credit customer data to be predicted, corresponding state tag data and
Customer relationship map；Risk of fraud detection module is used for credit customer data, corresponding state tag data and customer relationship
Map is input in credit fraud detection model, obtains the corresponding risk of fraud of credit customer.
5th aspect, the embodiment of the present invention also provide a kind of server, which includes memory and processor；Storage
Device supports processor perform claim to require credit fraud detection model training method or second described in first aspect for storing
The program of credit fraud detection method described in aspect, processor are configurable for executing the program stored in memory.
The embodiment of the present invention bring it is following the utility model has the advantages that
The present invention provides a kind of credit fraud detection method and its training method, device and the servers of model, wherein
In the training method of credit fraud detection model, client statistical nature vector sum client is determined by the credit customer data of acquisition
Relation map；By the corresponding state tag data of the credit customer data of acquisition, by the sequence node in customer relationship map
Be converted to node label sequence；Then by skip-gram algorithm, node label is Sequence Transformed for customer relationship map correspondence
Feature vector；According to the feature vector of statistical nature vector sum customer relationship map, initial model is trained, until instruction
Experienced the number of iterations meets preset the number of iterations threshold value, obtains credit fraud detection model.The credit fraud that the present invention obtains
Detection model takes full advantage of relation map network structure, label characteristics and statistical nature vector, which can preferably know
The risk of fraud of other client, to improve the reliability of risk supervision result.
Other features and advantages of the present invention will illustrate in the following description, alternatively, Partial Feature and advantage can be with
Deduce from specification or unambiguously determine, or by implementing above-mentioned technology of the invention it can be learnt that.
To enable the above objects, features and advantages of the present invention to be clearer and more comprehensible, better embodiment is cited below particularly, and match
Appended attached drawing is closed, is described in detail below.
Detailed description of the invention
It, below will be to specific in order to illustrate more clearly of the specific embodiment of the invention or technical solution in the prior art
Embodiment or attached drawing needed to be used in the description of the prior art be briefly described, it should be apparent that, it is described below
Attached drawing is some embodiments of the present invention, for those of ordinary skill in the art, before not making the creative labor
It puts, is also possible to obtain other drawings based on these drawings.
Fig. 1 is a kind of flow chart of the training method of credit fraud detection model provided in an embodiment of the present invention；
Fig. 2 is the schematic diagram of customer relationship map provided in an embodiment of the present invention；
Fig. 3 is the flow chart of the training method of another credit fraud detection model provided in an embodiment of the present invention；
Fig. 4 is a kind of flow chart of credit fraud detection method provided in an embodiment of the present invention；
Fig. 5 is a kind of structural schematic diagram of the training device of credit fraud detection model provided in an embodiment of the present invention；
Fig. 6 is a kind of structural schematic diagram of credit fraud detection device provided in an embodiment of the present invention.
Specific embodiment
In order to make the object, technical scheme and advantages of the embodiment of the invention clearer, below in conjunction with attached drawing to the present invention
Technical solution be clearly and completely described, it is clear that described embodiments are some of the embodiments of the present invention, rather than
Whole embodiments.Based on the embodiments of the present invention, those of ordinary skill in the art are not making creative work premise
Under every other embodiment obtained, shall fall within the protection scope of the present invention.
Rely only on statistics of attributes information in the prior art to detect credit fraud, the information source for detection is more single
One, lead to testing result poor reliability；With the development in Internet technology of relation map, many internet financial institutions
Relation map is applied to internet credit field, the network structure for being mainly based upon relation map carries out certain information characteristics
When extracting, but facing the complex network of substantial amounts, information characteristics extraction can have loss of learning, it is difficult to give full expression to relational graph
Therefore network structure in spectrum when data volume is huge, is still difficult to obtain reliable credit fraud detection knot according to relation map
Fruit.
Based on this, training method, the device of a kind of credit fraud detection method and its model provided in an embodiment of the present invention
And server, the technology can be applied to the risk of fraud of credit applications user under monitoring internet credit scene.
For convenient for understanding the present embodiment, first to a kind of credit fraud detection mould disclosed in the embodiment of the present invention
The training method of type describes in detail.
A kind of flow chart of the training method of credit fraud detection model shown in Figure 1, the credit fraud detection mould
The training method of type includes the following steps:
Step S102 obtains credit customer data and corresponding state tag data；
On the basis of credit customer authorization, above-mentioned credit is obtained by credit APP (Application, application program)
Customer data；The credit customer data include device data, request for data and social data；Device data includes customer equipment
IP (Internet Protocol, Internet protocol) address, MAC (Media Access Control, media access control)
The hardware such as location and IMEI (International Mobile Equipment Identity, international mobile equipment identification number) are set
Standby data；Request for data includes the Shens such as cell-phone number, identification card number, home address, CompanyAddress and the Business Name of client's application
It please data information；Social data includes the social informations such as contact person and the call of client.
The corresponding state tag data of above-mentioned credit customer are obtained again by credit APP, the state tag number
According to include client whether cheat, settle a claim, overdue, the service condition label closing and normally refund.
Step S104 determines client's statistical nature vector sum customer relationship map according to credit customer data；
Above-mentioned customer relationship map includes request for data, attribute data and corresponding customer relationship；Above-mentioned attribute data is
Data of the client obtained by device data in each application node, including company's node and addressed nodes etc.；Corresponding visitor
Family relationship includes colleague, friend, spouse, lineal relative, firm telephone, CompanyAddress and home address etc., and customer relationship
Different types of relationship is provided with different relationship weights by map, so that the feature of the subsequent customer relationship map being calculated
Vector is more accurate.
Above-mentioned client's statistical nature vector includes behavioral statistics feature vector, application statistical nature vector sum relationship statistics spy
Vector is levied, client's statistical nature vector is obtained by following manner:
Obtain device data, request for data and the social data of credit customer data；By the calculating to device data, obtain
To behavioral statistics feature vector；By the calculating to request for data, application statistical nature vector is obtained；By to social data
It calculates, obtains relationship statistical nature vector.
Statistics calculating is carried out to the device data, request for data and social data of credit customer data respectively, it is available
Corresponding behavioral statistics feature vector, application statistical nature vector sum relationship statistical nature vector.
Step S106, by the corresponding state tag data of credit customer data, by the corresponding client of credit customer data
Sequence node in relation map is converted to node label sequence；
According to the relationship weight in above-mentioned customer relationship map, and node2vec algorithm is combined, above-mentioned client can be obtained and close
It is the sequence node of map；Random walk path of the above-mentioned node2vec algorithm usually using the node of construction on network, mould
The process of imitative text generation, to obtain sequence node.
The sequence node includes application node and attribute node, and application node includes the status information of client, the business
Status information can be the service condition label for cheating, settling a claim, is overdue, closing and normally refund；Attribute node includes user's
Device data, request for data and social data, include figure and number in the sequence node in relation map, usually above-mentioned
Different information is also indicated with different figure and data in application node and attribute node；It is illustrated in figure 2 customer relational graph
The schematic diagram of spectrum, in Fig. 2, circle represents application node, and diamond shape represents attribute node, then the customer relationship map in Fig. 2
Sequence node is represented by [Isosorbide-5-Nitrae, 5,7,10].
Sequence node in the corresponding customer relationship map of credit customer data is converted into node label sequence, is equivalent to
By the number in the status information replacement sequence node in state tag data, for example, sequence node [Isosorbide-5-Nitrae, 5,7,10] is replaced
It is changed to node label sequence [normal, spouse is overdue, works together, fraud].
Step S108, it is by skip-gram algorithm, node label is Sequence Transformed for the corresponding feature of customer relationship map
Vector；
Skip-gram algorithm can predict context according to input word, and skip-gram algorithm is usually by node label sequence
Column are input in skip-gram model, which can export multidimensional related with the data according to the input data
Feature vector (is equivalent to multiple sequences), which is the corresponding feature vector of customer relationship map.
Step S110 is trained initial model according to the feature vector of statistical nature vector sum customer relationship map,
Until the number of iterations of training meets preset the number of iterations threshold value, credit fraud detection model is obtained.
Above-mentioned initial model is usually preset model, and the initial model by a large amount of data for being instructed
Practice, obtains ideal credit and its detection model.
The feature vector of statistical nature vector sum relation map is merged, initial model is trained；In model
Need to preset trained the number of iterations threshold value before training, which is to utilize Grid Search grid search
What method obtained, which is usually a kind of tune ginseng means, and basic thought is all candidate
In parameter selection, by looping through, each possibility is attempted, using the parameter to behave oneself best as final parameter；The party
The number of iterations threshold value (being equivalent to above-mentioned final parameter) that method obtains, available reliable threshold value are conducive to improve training
The reliability of model.
The present invention provides the credit customer numbers that a kind of training method of credit fraud detection model, this method pass through acquisition
According to determining client's statistical nature vector sum customer relationship map；Pass through the corresponding state tag number of the credit customer data of acquisition
According to the sequence node in customer relationship map is converted to node label sequence；Then by skip-gram algorithm, by node
Sequence label is converted into the corresponding feature vector of customer relationship map；According to the feature of statistical nature vector sum customer relationship map
Vector is trained initial model, until the number of iterations of training meets preset the number of iterations threshold value, obtains credit fraud
Detection model.The credit fraud detection model that the present invention obtains takes full advantage of relation map network structure, label characteristics and system
Feature vector is counted, which can preferably identify the risk of fraud of client, to improve the reliability of risk supervision result.
The flow chart of the training method of another credit fraud detection model shown in Figure 3, this method is in above-mentioned Fig. 1
It is realized on the basis of method provided by illustrated embodiment；This method comprises the following steps:
Step S302 obtains credit customer data and corresponding state tag data；
Step S304 determines client's statistical nature vector sum customer relationship map according to credit customer data；
Step S306 obtains customer relationship map by the preset relationship weight of customer relationship and Random Walk Algorithm
Sequence node；The sequence node includes application node and attribute node.
The above-mentioned preset relationship weight of customer relationship is the weight that user is arranged according to customer relationship in customer relationship map,
Wherein different customer relationships is provided with different weights, for example, customer relationship is more intimate, the weight of setting is bigger；Client's connection
System is closer, and the weight of setting is bigger etc..
Above-mentioned Random Walk Algorithm is Random Walk Algorithm in node2vec algorithm, and node2vec algorithm is by introducing two
Breadth-first search (BFS) and depth-first search (DFS) are introduced into the generating process of random walk sequence by parameter p and q；
What wherein parameter p and q was used to control random walk sequence jumps probability.
The process of above-mentioned formation sequence node uses network representation and learns (network representation
Learning) technology, by network low-dimensional vector (being equivalent to above-mentioned sequence node) indicate, by by large-scale complex network to
Quantization means can be analyzed and be modeled to modeling methods such as Web vector graphic cluster, classification；Network representation learning method
Node2vec combines BFS and DFS by introducing the sequence generation strategy being biased on the basis of Random Walk Algorithm
Two ways carries out vectorization expression to network, can preferably retain network structure, and then obtain more satisfactory node
Sequence.
The random walk sequence obtained by above-mentioned relation weight combination Random Walk Algorithm, available customer relational graph
The sequence node of spectrum, sequence node are the sequences being made of a string of node IDs (identity, identity number)；Each node
ID represents different credit applications information, and node includes application node and the associated attribute node of application node.
Step S308 converts the application node in the sequence node in the corresponding customer relationship map of credit customer data
For the corresponding state tag data of credit customer data；
Corresponding state tag is converted by the node ID in application node, for example, converting application node " 1 " to " just
Often " (it is normal for being equivalent to credit customer service condition)；" overdue ", which is converted, by application node " 5 " (is equivalent to credit customer industry
Business state is overdue)；Wherein, the transforming relationship of the node ID of application node and state tag data is pre-set.
Step S310 converts the attribute node in the sequence node in the corresponding customer relationship map of credit customer data
For request for data, attribute data or the corresponding customer relationship in customer relationship map, to obtain the corresponding section of sequence node
Point sequence label；
Attribute type is converted by attribute node, which is equivalent to the application number in above-mentioned customer relationship map
According to, attribute data or corresponding customer relationship, comprising: the IP address of user equipment, MAC Address, IMEI, client apply for mobile phone
Number, identification card number, home address, CompanyAddress, Business Name or address book contact etc..
The corresponding state tag data of credit customer data are converted by above-mentioned application node, attribute node is converted into client
It after request for data, attribute data or corresponding customer relationship in relation map, is integrated, available corresponding node
Sequence label.
Step S312, by node label sequence inputting to skip-gram model, output obtains the feature of customer relationship map
Vector；
Above-mentioned skip-gram model can predict context according to input time, usually extremely by node label sequence inputting
In the model, which can export multidimensional characteristic vectors related with the data according to the input data and (be equivalent to multiple sequences
Column), which is the feature vector of customer relationship map.
Step S314 merges the feature vector of statistical nature vector sum customer relationship map；
Step S316 is trained initial model using supervised classification algorithm, works as instruction according to the feature vector after merging
When experienced the number of iterations meets preset the number of iterations threshold value, credit fraud detection model is obtained.
By the behavioral statistics feature vector of client's statistical nature vector, application statistical nature vector sum relationship statistical nature to
It measures and is merged with the feature vector of customer relationship map；It is first using the feature vector training after merging based on supervised classification algorithm
Beginning model obtains credit fraud detection model when trained the number of iterations meets preset the number of iterations threshold value.
Above-mentioned supervised classification algorithm is also known as coaching method, is to establish statistics recognition function as theoretical basis, foundation typical sample
The technology that this training method is classified；The algorithm can be found out according to the feature vector after merging by selecting characteristic parameter
Characteristic parameter establishes discriminant function and classifies to feature vector to be sorted as decision rule；It can be obtained according to classification results
To training pattern.
The embodiment of the present invention has merged relation map network structure feature and label characteristics, and the feature for generating vectorization is used
In the identification of risk of fraud, while remaining network structure information, also fused business label information, can will close this method
It is that the network information and business attribute information in map is fully utilized, so that credit fraud detection model can be better
Identify the risk of fraud of client.
Corresponding to the training method of above-mentioned credit fraud detection model, the embodiment of the present invention also provides a kind of credit fraud inspection
Survey method, this method are applied to the equipment configured with credit fraud detection model；The credit fraud detection model is above-mentioned implementation
The model that the training method training of credit fraud detection model in example obtains；As described in Figure 4, this method comprises the following steps:
Step S402 obtains credit customer data to be predicted, corresponding state tag data and customer relationship map；
The credit customer data to be predicted include device data, request for data and social data；The state tag data
State tag information including credit customer moment before current time, the state tag information include cheating, settling a claim, exceeding
Phase such as closes and normally refunds at the states；The state tag data further include other visitors associated with credit customer to be predicted
The state tag information at family.
Credit customer data, state tag data and customer relationship map are input to credit fraud detection by step S404
In model, the corresponding risk of fraud of the credit customer is obtained.
Above-mentioned credit fraud detection model can be according to the credit customer data, corresponding state tag data and client of input
Relation map is adjusted, and obtains credit fraud model corresponding with the credit customer, the risk of fraud obtained in this way is more
Accurately.
Above-mentioned credit fraud detection method obtains credit customer data, corresponding state tag data and customer relational graph
After spectrum, which is input in credit fraud detection model, corresponding risk of fraud can be obtained.Which passes through adjustable
Credit fraud detection model, enables model preferably to identify the risk of fraud of credit customer, to be the wealth of credit customer
It produces and preferably protection is provided.
Corresponding to the training method embodiment of above-mentioned credit fraud detection model, a kind of credit fraud shown in Figure 5
The structural schematic diagram of the training device of detection model；The device includes:
Customer data obtains module 50, for obtaining credit customer data and corresponding state tag data；
Map construction module 51, for determining client's statistical nature vector sum customer relational graph according to credit customer data
Spectrum；
Sequence conversion module 52, for passing through the corresponding state tag data of credit customer data, by credit customer data
Sequence node in corresponding customer relationship map is converted to node label sequence；
Feature vector module 53, for node label sequence inputting into skip-gram model, to be exported customer relationship
The feature vector of map；
Model training module 54, for the feature vector according to statistical nature vector sum customer relationship map, using supervision
Sorting algorithm carries out model training, until trained the number of iterations meets preset the number of iterations threshold value, obtains credit fraud inspection
Survey model.
The training device of a kind of credit fraud detection model provided in an embodiment of the present invention, with provided by the above embodiment one
The training method technical characteristic having the same of kind credit fraud detection model reaches so also can solve identical technical problem
To identical technical effect.
Corresponding to the embodiment of above-mentioned credit fraud detection method, a kind of credit fraud detection device shown in Figure 6
Structural schematic diagram；The device is applied to the equipment configured with credit fraud detection model；The credit fraud detection model is upper
State the model that the training method of credit fraud detection module obtains；The device includes:
Data acquisition module 60, for obtaining the application of credit customer and other associated credit customers to be predicted
Data and corresponding state tag data；
Risk of fraud detection module 61, for credit customer data and state tag data to be input to credit fraud detection
In model, credit customer data and the corresponding risk of fraud of state tag data are obtained.
A kind of credit fraud detection device provided in an embodiment of the present invention, with a kind of credit fraud provided by the above embodiment
Detection method technical characteristic having the same reaches identical technical effect so also can solve identical technical problem.
The present embodiment additionally provides a kind of a kind of server corresponding to the above method embodiment, which includes depositing
Reservoir and processor；Memory supports processor to execute credit fraud detection model training method or credit fraud for storing
The program of detection method, processor are configurable for executing the program stored in memory.
The training method of credit fraud detection method and its model, device provided by the embodiment of the present invention and server
Computer program product, the computer readable storage medium including storing program code, the instruction that said program code includes
It can be used for executing previous methods method as described in the examples, specific implementation can be found in embodiment of the method, and details are not described herein.
The training method of credit fraud detection method and its model, device provided by the embodiment of the present invention and server,
Vectorization is carried out by the relation map fusion tag attribute information generated to credit applications client, and feature vector is calculated, and
It is obtained in conjunction with other attribute feature vectors building model with detecting the risk of fraud of credit applications client under internet credit scene
The accuracy rate of the risk of fraud arrived is higher.
It is apparent to those skilled in the art that for convenience and simplicity of description, the service of foregoing description
The specific work process of device, can refer to corresponding processes in the foregoing method embodiment, and details are not described herein.
It, can be with if the function is realized in the form of SFU software functional unit and when sold or used as an independent product
It is stored in a computer readable storage medium.Based on this understanding, technical solution of the present invention is substantially in other words
The part of the part that contributes to existing technology or the technical solution can be embodied in the form of software products, the meter
Calculation machine software product is stored in a storage medium, including some instructions are used so that a computer equipment (can be a
People's computer, server or network equipment etc.) it performs all or part of the steps of the method described in the various embodiments of the present invention.
And storage medium above-mentioned includes: that USB flash disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), arbitrary access are deposited
The various media that can store program code such as reservoir (RAM, Random Access Memory), magnetic or disk.
Finally, it should be noted that embodiment described above, only a specific embodiment of the invention, to illustrate the present invention
Technical solution, rather than its limitations, scope of protection of the present invention is not limited thereto, although with reference to the foregoing embodiments to this hair
It is bright to be described in detail, those skilled in the art should understand that: anyone skilled in the art
In the technical scope disclosed by the present invention, it can still modify to technical solution documented by previous embodiment or can be light
It is readily conceivable that variation or equivalent replacement of some of the technical features；And these modifications, variation or replacement, do not make
The essence of corresponding technical solution is detached from the spirit and scope of technical solution of the embodiment of the present invention, should all cover in protection of the invention
Within the scope of.Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.
Claims (10)
1. a kind of training method of credit fraud detection model, which is characterized in that the described method includes:
Obtain credit customer data and corresponding state tag data；
According to the credit customer data, client's statistical nature vector sum customer relationship map is determined；
By the corresponding state tag data of the credit customer data, by the corresponding visitor of the credit customer data
Sequence node in the relation map of family is converted to node label sequence；
It is by skip-gram algorithm, the node label is Sequence Transformed for the corresponding feature vector of the customer relationship map；
According to the feature vector of customer relationship map described in the statistical nature vector sum, initial model is trained, until
Trained the number of iterations meets preset the number of iterations threshold value, obtains credit fraud detection model.
2. the method according to claim 1, wherein the customer relationship map includes request for data, attribute number
According to corresponding customer relationship；Client's statistical nature vector includes behavioral statistics feature vector, application statistical nature vector
With relationship statistical nature vector, client's statistical nature vector is obtained by following manner:
Obtain device data, request for data and the social data of the credit customer data；
By the calculating to the device data, the behavioral statistics feature vector is obtained；
By the calculating to the request for data, the application statistical nature vector is obtained；
By the calculating to the social data, the relationship statistical nature vector is obtained.
3. the method according to claim 1, wherein passing through the corresponding state of the credit customer data
Sequence node in the corresponding customer relationship map of the credit customer data is converted to node label sequence by label data
Before the step of column, further includes:
By the preset relationship weight of customer relationship and Random Walk Algorithm, the sequence node of the customer relationship map is obtained；
The sequence node includes application node and attribute node.
4. the method according to claim 1, wherein described pass through the corresponding shape of the credit customer data
Sequence node in the corresponding customer relationship map of the credit customer data is converted to node label by state label data
The step of sequence includes:
Institute is converted by the application node in the sequence node in the corresponding customer relationship map of the credit customer data
State the corresponding state tag data of credit customer data；
Visitor is converted by the attribute node in the sequence node in the corresponding customer relationship map of the credit customer data
Request for data, attribute data or corresponding customer relationship in the relation map of family, to obtain the corresponding node of the sequence node
Sequence label.
5. the method according to claim 1, wherein described by skip-gram algorithm, by the node label
The step of Sequence Transformed feature vector corresponding for the customer relationship map, comprising:
By the node label sequence inputting to skip-gram model, output obtains the corresponding feature of the customer relationship map
Vector.
6. the method according to claim 1, wherein the client according to the statistical nature vector sum is closed
It is the feature vector of map, the initial model is trained, until the number of iterations of training meets preset the number of iterations
Threshold value, the step of obtaining credit fraud detection model include:
The feature vector of customer relationship map described in the statistical nature vector sum is merged；
According to the feature vector after merging, the initial model is trained using supervised classification algorithm, when trained iteration
When number meets preset the number of iterations threshold value, credit fraud detection model is obtained.
7. a kind of credit fraud detection method, which is characterized in that the method is applied to configured with credit fraud detection model
Equipment；The credit fraud detection model is the model that the method training of any one of claim 1 to 6 obtains；The method
Include:
Obtain credit customer data to be predicted, corresponding state tag data and customer relationship map；
The credit customer data, the state tag data and the customer relationship map are input to the credit fraud inspection
It surveys in model, obtains the corresponding risk of fraud of credit customer.
8. a kind of training device of credit fraud detection model, which is characterized in that described device includes:
Customer data obtains module, for obtaining credit customer data and corresponding state tag data；
Map construction module, for determining client's statistical nature vector sum customer relationship map according to the credit customer data；
Sequence conversion module is used for through the corresponding state tag data of the credit customer data, by the credit visitor
Sequence node in the corresponding customer relationship map of user data is converted to node label sequence；
Feature vector module, it is for passing through skip-gram algorithm, the node label is Sequence Transformed for the customer relationship
The corresponding feature vector of map；
Model training module, for the feature vector of the customer relationship map according to the statistical nature vector sum, to initial
Model is trained, and until trained the number of iterations meets preset the number of iterations threshold value, obtains credit fraud detection model.
9. a kind of credit fraud detection device, which is characterized in that described device is applied to configured with credit fraud detection model
Equipment；The credit fraud detection model is the model that the method training of any one of claim 1 to 6 obtains；Described device
Include:
Data acquisition module, for obtaining credit customer data to be predicted, corresponding state tag data and customer relational graph
Spectrum；
Risk of fraud detection module is used for the credit customer data, the state tag data and the customer relational graph
Spectrum is input in the credit fraud detection model, obtains the corresponding risk of fraud of credit customer.
10. a kind of server, which is characterized in that the server includes memory and processor；The memory is for storing
Processor perform claim is supported to require any one of 1 to the 6 credit fraud detection model training method or claim 7 institute
The program for the credit fraud detection method stated, the processor is configured to for executing the program stored in the memory.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811616968.1A CN109685647B (en) | 2018-12-27 | 2018-12-27 | Credit fraud detection method and training method and device of model thereof, and server |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811616968.1A CN109685647B (en) | 2018-12-27 | 2018-12-27 | Credit fraud detection method and training method and device of model thereof, and server |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109685647A true CN109685647A (en) | 2019-04-26 |
| CN109685647B CN109685647B (en) | 2021-08-10 |
Family
ID=66190619
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811616968.1A Active CN109685647B (en) | 2018-12-27 | 2018-12-27 | Credit fraud detection method and training method and device of model thereof, and server |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109685647B (en) |
Cited By (26)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110210227A (en) * | 2019-06-11 | 2019-09-06 | 百度在线网络技术（北京）有限公司 | Risk checking method, device, equipment and storage medium |
| CN110223168A (en) * | 2019-06-24 | 2019-09-10 | 浪潮卓数大数据产业发展有限公司 | A kind of anti-fraud detection method of label propagation and system based on business connection map |
| CN110688536A (en) * | 2019-09-25 | 2020-01-14 | 中国建设银行股份有限公司 | Label prediction method, device, equipment and storage medium |
| CN110717823A (en) * | 2019-09-29 | 2020-01-21 | 支付宝(杭州)信息技术有限公司 | Credit overdue risk identification method and system |
| CN110751557A (en) * | 2019-10-10 | 2020-02-04 | 中国建设银行股份有限公司 | Abnormal fund transaction behavior analysis method and system based on sequence model |
| CN110765117A (en) * | 2019-09-30 | 2020-02-07 | 中国建设银行股份有限公司 | Fraud identification method and device, electronic equipment and computer-readable storage medium |
| CN110796269A (en) * | 2019-09-30 | 2020-02-14 | 北京明略软件系统有限公司 | Method and device for generating model, and method and device for processing information |
| CN111104521A (en) * | 2019-12-18 | 2020-05-05 | 上海观安信息技术股份有限公司 | Anti-fraud detection method and detection system based on graph analysis |
| CN111127185A (en) * | 2019-11-25 | 2020-05-08 | 北京明略软件系统有限公司 | Credit fraud identification model construction method and device |
| CN111143430A (en) * | 2019-12-06 | 2020-05-12 | 北京明略软件系统有限公司 | Guarantee data mining method and system |
| CN111178615A (en) * | 2019-12-24 | 2020-05-19 | 成都数联铭品科技有限公司 | Construction method and system of enterprise risk identification model |
| CN111198967A (en) * | 2019-12-20 | 2020-05-26 | 北京淇瑀信息科技有限公司 | User grouping method and device based on relational graph and electronic equipment |
| CN111222981A (en) * | 2020-01-16 | 2020-06-02 | 中国建设银行股份有限公司 | Credibility determination method, device, equipment and storage medium |
| CN111641594A (en) * | 2020-05-09 | 2020-09-08 | 同济大学 | Method, system, medium and device for detecting fraudulent user based on page behavior |
| CN111694969A (en) * | 2020-06-18 | 2020-09-22 | 拉卡拉支付股份有限公司 | User identity identification method and device |
| CN111861702A (en) * | 2020-07-09 | 2020-10-30 | 睿智合创(北京)科技有限公司 | Case qualitative method and system based on anti-fraud variable and decision result |
| CN112036165A (en) * | 2019-05-14 | 2020-12-04 | 西交利物浦大学 | Method for constructing news characteristic vector and application |
| CN112258315A (en) * | 2020-10-20 | 2021-01-22 | 南京三百云信息科技有限公司 | Vehicle loan pre-loan data inspection method and device based on identity tag |
| WO2021012913A1 (en) * | 2019-07-23 | 2021-01-28 | 平安科技（深圳）有限公司 | Data recognition method and system, electronic device and computer storage medium |
| CN113610122A (en) * | 2021-07-22 | 2021-11-05 | 上海淇玥信息技术有限公司 | User equipment authentication method and device and computer equipment |
| CN113706291A (en) * | 2021-08-31 | 2021-11-26 | 平安普惠企业管理有限公司 | Fraud risk prediction method, device, equipment and storage medium |
| CN114169458A (en) * | 2021-12-13 | 2022-03-11 | 平安科技（深圳）有限公司 | Method and device for identifying cheater, storage medium and computer equipment |
| CN114970548A (en) * | 2022-06-14 | 2022-08-30 | 阿里云计算有限公司 | Label processing method and device, electronic equipment and computer readable storage medium |
| CN115034520A (en) * | 2022-08-09 | 2022-09-09 | 太平金融科技服务(上海)有限公司深圳分公司 | Risk prediction method, device, equipment and storage medium |
| CN110689423B (en) * | 2019-08-22 | 2023-08-11 | 平安科技（深圳）有限公司 | Credit evaluation method and device |
| CN117035082A (en) * | 2023-10-10 | 2023-11-10 | 北京江融信科技有限公司 | Knowledge graph real-time operation method and system for identifying fraudulent party |
Citations (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106445988A (en) * | 2016-06-01 | 2017-02-22 | 上海坤士合生信息科技有限公司 | Intelligent big data processing method and system |
| CN106997474A (en) * | 2016-12-29 | 2017-08-01 | 南京邮电大学 | A kind of node of graph multi-tag sorting technique based on deep learning |
| CN107169086A (en) * | 2017-05-12 | 2017-09-15 | 北京化工大学 | A kind of file classification method |
| CN107169864A (en) * | 2017-05-31 | 2017-09-15 | 天云融创数据科技（北京）有限公司 | A kind of card holder's risk of fraud feature extracting method based on complex network |
| CN107194803A (en) * | 2017-05-19 | 2017-09-22 | 南京工业大学 | P2P net loan borrower credit risk assessment device |
| CN107862053A (en) * | 2017-11-08 | 2018-03-30 | 北京奇虎科技有限公司 | User's portrait building method, device and computing device based on customer relationship |
| CN108256907A (en) * | 2018-01-09 | 2018-07-06 | 北京腾云天下科技有限公司 | A kind of construction method and computing device of customer grouping model |
| CN108399509A (en) * | 2018-04-12 | 2018-08-14 | 阿里巴巴集团控股有限公司 | Determine the method and device of the risk probability of service request event |
| CN108510280A (en) * | 2018-03-23 | 2018-09-07 | 上海氪信信息技术有限公司 | A kind of financial fraud behavior prediction method based on mobile device behavioral data |
| CN108564460A (en) * | 2018-01-12 | 2018-09-21 | 阳光财产保险股份有限公司 | Real-time fraud detection method under internet credit scene and device |
| CN108647800A (en) * | 2018-03-19 | 2018-10-12 | 浙江工业大学 | A kind of online social network user missing attribute forecast method based on node insertion |
| CN108765131A (en) * | 2018-05-18 | 2018-11-06 | 深圳壹账通智能科技有限公司 | Credit authorization method, apparatus, terminal and readable storage medium storing program for executing based on micro- expression |
| CN109035003A (en) * | 2018-07-04 | 2018-12-18 | 北京玖富普惠信息技术有限公司 | Anti- fraud model modelling approach and anti-fraud monitoring method based on machine learning |
-
2018
- 2018-12-27 CN CN201811616968.1A patent/CN109685647B/en active Active
Patent Citations (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106445988A (en) * | 2016-06-01 | 2017-02-22 | 上海坤士合生信息科技有限公司 | Intelligent big data processing method and system |
| CN106997474A (en) * | 2016-12-29 | 2017-08-01 | 南京邮电大学 | A kind of node of graph multi-tag sorting technique based on deep learning |
| CN107169086A (en) * | 2017-05-12 | 2017-09-15 | 北京化工大学 | A kind of file classification method |
| CN107194803A (en) * | 2017-05-19 | 2017-09-22 | 南京工业大学 | P2P net loan borrower credit risk assessment device |
| CN107169864A (en) * | 2017-05-31 | 2017-09-15 | 天云融创数据科技（北京）有限公司 | A kind of card holder's risk of fraud feature extracting method based on complex network |
| CN107862053A (en) * | 2017-11-08 | 2018-03-30 | 北京奇虎科技有限公司 | User's portrait building method, device and computing device based on customer relationship |
| CN108256907A (en) * | 2018-01-09 | 2018-07-06 | 北京腾云天下科技有限公司 | A kind of construction method and computing device of customer grouping model |
| CN108564460A (en) * | 2018-01-12 | 2018-09-21 | 阳光财产保险股份有限公司 | Real-time fraud detection method under internet credit scene and device |
| CN108647800A (en) * | 2018-03-19 | 2018-10-12 | 浙江工业大学 | A kind of online social network user missing attribute forecast method based on node insertion |
| CN108510280A (en) * | 2018-03-23 | 2018-09-07 | 上海氪信信息技术有限公司 | A kind of financial fraud behavior prediction method based on mobile device behavioral data |
| CN108399509A (en) * | 2018-04-12 | 2018-08-14 | 阿里巴巴集团控股有限公司 | Determine the method and device of the risk probability of service request event |
| CN108765131A (en) * | 2018-05-18 | 2018-11-06 | 深圳壹账通智能科技有限公司 | Credit authorization method, apparatus, terminal and readable storage medium storing program for executing based on micro- expression |
| CN109035003A (en) * | 2018-07-04 | 2018-12-18 | 北京玖富普惠信息技术有限公司 | Anti- fraud model modelling approach and anti-fraud monitoring method based on machine learning |
Non-Patent Citations (1)
| Title |
|---|
| 冯曦等: ""基于深度游走模型的标签传播社区发现算法"", 《计算机工程》 * |
Cited By (38)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112036165A (en) * | 2019-05-14 | 2020-12-04 | 西交利物浦大学 | Method for constructing news characteristic vector and application |
| CN110210227B (en) * | 2019-06-11 | 2021-05-14 | 百度在线网络技术（北京）有限公司 | Risk detection method, device, equipment and storage medium |
| CN110210227A (en) * | 2019-06-11 | 2019-09-06 | 百度在线网络技术（北京）有限公司 | Risk checking method, device, equipment and storage medium |
| CN110223168A (en) * | 2019-06-24 | 2019-09-10 | 浪潮卓数大数据产业发展有限公司 | A kind of anti-fraud detection method of label propagation and system based on business connection map |
| CN110223168B (en) * | 2019-06-24 | 2022-06-28 | 浪潮卓数大数据产业发展有限公司 | Label propagation anti-fraud detection method and system based on enterprise relationship map |
| WO2021012913A1 (en) * | 2019-07-23 | 2021-01-28 | 平安科技（深圳）有限公司 | Data recognition method and system, electronic device and computer storage medium |
| CN110689423B (en) * | 2019-08-22 | 2023-08-11 | 平安科技（深圳）有限公司 | Credit evaluation method and device |
| CN110688536A (en) * | 2019-09-25 | 2020-01-14 | 中国建设银行股份有限公司 | Label prediction method, device, equipment and storage medium |
| CN110717823A (en) * | 2019-09-29 | 2020-01-21 | 支付宝(杭州)信息技术有限公司 | Credit overdue risk identification method and system |
| CN110717823B (en) * | 2019-09-29 | 2022-08-02 | 支付宝(杭州)信息技术有限公司 | Credit overdue risk identification method and system |
| CN110765117B (en) * | 2019-09-30 | 2023-09-26 | 建信金融科技有限责任公司 | Fraud identification method, fraud identification device, electronic equipment and computer readable storage medium |
| CN110796269A (en) * | 2019-09-30 | 2020-02-14 | 北京明略软件系统有限公司 | Method and device for generating model, and method and device for processing information |
| CN110796269B (en) * | 2019-09-30 | 2023-04-18 | 北京明略软件系统有限公司 | Method and device for generating model, and method and device for processing information |
| CN110765117A (en) * | 2019-09-30 | 2020-02-07 | 中国建设银行股份有限公司 | Fraud identification method and device, electronic equipment and computer-readable storage medium |
| CN110751557B (en) * | 2019-10-10 | 2023-04-18 | 建信金融科技有限责任公司 | Abnormal fund transaction behavior analysis method and system based on sequence model |
| CN110751557A (en) * | 2019-10-10 | 2020-02-04 | 中国建设银行股份有限公司 | Abnormal fund transaction behavior analysis method and system based on sequence model |
| CN111127185A (en) * | 2019-11-25 | 2020-05-08 | 北京明略软件系统有限公司 | Credit fraud identification model construction method and device |
| CN111143430A (en) * | 2019-12-06 | 2020-05-12 | 北京明略软件系统有限公司 | Guarantee data mining method and system |
| CN111104521A (en) * | 2019-12-18 | 2020-05-05 | 上海观安信息技术股份有限公司 | Anti-fraud detection method and detection system based on graph analysis |
| CN111104521B (en) * | 2019-12-18 | 2023-10-17 | 上海观安信息技术股份有限公司 | Anti-fraud detection method and detection system based on graph analysis |
| CN111198967A (en) * | 2019-12-20 | 2020-05-26 | 北京淇瑀信息科技有限公司 | User grouping method and device based on relational graph and electronic equipment |
| CN111198967B (en) * | 2019-12-20 | 2024-03-08 | 北京淇瑀信息科技有限公司 | User grouping method and device based on relationship graph and electronic equipment |
| CN111178615A (en) * | 2019-12-24 | 2020-05-19 | 成都数联铭品科技有限公司 | Construction method and system of enterprise risk identification model |
| CN111178615B (en) * | 2019-12-24 | 2023-10-27 | 成都数联铭品科技有限公司 | Method and system for constructing enterprise risk identification model |
| CN111222981A (en) * | 2020-01-16 | 2020-06-02 | 中国建设银行股份有限公司 | Credibility determination method, device, equipment and storage medium |
| CN111641594A (en) * | 2020-05-09 | 2020-09-08 | 同济大学 | Method, system, medium and device for detecting fraudulent user based on page behavior |
| CN111694969B (en) * | 2020-06-18 | 2021-07-09 | 拉卡拉支付股份有限公司 | User identity identification method and device |
| CN111694969A (en) * | 2020-06-18 | 2020-09-22 | 拉卡拉支付股份有限公司 | User identity identification method and device |
| CN111861702A (en) * | 2020-07-09 | 2020-10-30 | 睿智合创(北京)科技有限公司 | Case qualitative method and system based on anti-fraud variable and decision result |
| CN112258315B (en) * | 2020-10-20 | 2024-02-02 | 南京三百云信息科技有限公司 | Method and device for checking vehicle credit pre-credit data based on identity tag |
| CN112258315A (en) * | 2020-10-20 | 2021-01-22 | 南京三百云信息科技有限公司 | Vehicle loan pre-loan data inspection method and device based on identity tag |
| CN113610122A (en) * | 2021-07-22 | 2021-11-05 | 上海淇玥信息技术有限公司 | User equipment authentication method and device and computer equipment |
| CN113706291A (en) * | 2021-08-31 | 2021-11-26 | 平安普惠企业管理有限公司 | Fraud risk prediction method, device, equipment and storage medium |
| CN114169458A (en) * | 2021-12-13 | 2022-03-11 | 平安科技（深圳）有限公司 | Method and device for identifying cheater, storage medium and computer equipment |
| CN114970548A (en) * | 2022-06-14 | 2022-08-30 | 阿里云计算有限公司 | Label processing method and device, electronic equipment and computer readable storage medium |
| CN115034520B (en) * | 2022-08-09 | 2023-01-10 | 太平金融科技服务(上海)有限公司深圳分公司 | Risk prediction method, device, equipment and storage medium |
| CN115034520A (en) * | 2022-08-09 | 2022-09-09 | 太平金融科技服务(上海)有限公司深圳分公司 | Risk prediction method, device, equipment and storage medium |
| CN117035082A (en) * | 2023-10-10 | 2023-11-10 | 北京江融信科技有限公司 | Knowledge graph real-time operation method and system for identifying fraudulent party |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109685647B (en) | 2021-08-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109685647A (en) | The training method of credit fraud detection method and its model, device and server | |
| CN110223168B (en) | Label propagation anti-fraud detection method and system based on enterprise relationship map | |
| CN108564460B (en) | Real-time fraud detection method and device in internet credit scene | |
| CN110413707A (en) | The excavation of clique's relationship is cheated in internet and checks method and its system | |
| CN109214191A (en) | A method of utilizing deep learning forecasting software security breaches | |
| CN109525595A (en) | A kind of black production account recognition methods and equipment based on time flow feature | |
| CN105184160B (en) | A kind of method of the Android phone platform application program malicious act detection based on API object reference relational graphs | |
| CN110198310A (en) | A kind of anti-cheat method of network behavior, device and storage medium | |
| CN106021377A (en) | Information processing method and device implemented by computer | |
| CN111008337B (en) | Deep attention rumor identification method and device based on ternary characteristics | |
| CN109903053B (en) | Anti-fraud method for behavior recognition based on sensor data | |
| CN111159387A (en) | Recommendation method based on multi-dimensional alarm information text similarity analysis | |
| CN114202336A (en) | Risk behavior monitoring method and system in financial scene | |
| CN108062538A (en) | Face identification method and device | |
| CN109685643A (en) | Loan audit risk grade determines method, apparatus, equipment and storage medium | |
| CN110033284A (en) | Source of houses verification method, apparatus, equipment and storage medium | |
| CN109472626A (en) | A kind of intelligent finance risk control method and system towards mobile phone charter business | |
| CN109213831A (en) | Event detecting method and device calculate equipment and storage medium | |
| CN110008828A (en) | Pairs of constraint ingredient assay measures optimization method based on difference regularization | |
| CN115186012A (en) | Power consumption data detection method, device, equipment and storage medium | |
| CN107402878A (en) | Method of testing and device | |
| CN114140248A (en) | AI artificial intelligence technology-based abnormal transaction identification method | |
| CN115619245A (en) | Portrait construction and classification method and system based on data dimension reduction method | |
| CN109088744A (en) | Powerline network abnormal intrusion detection method, device, equipment and storage medium | |
| CN112686654A (en) | Block chain digital currency transaction identification 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 |