CN108564460A - Real-time fraud detection method under internet credit scene and device - Google Patents
Real-time fraud detection method under internet credit scene and device Download PDFInfo
- Publication number
- CN108564460A CN108564460A CN201810033615.2A CN201810033615A CN108564460A CN 108564460 A CN108564460 A CN 108564460A CN 201810033615 A CN201810033615 A CN 201810033615A CN 108564460 A CN108564460 A CN 108564460A
- Authority
- CN
- China
- Prior art keywords
- target user
- data
- fraud
- target
- vector
- 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
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
-
- 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
Abstract
The present invention provides under a kind of internet credit scene real-time fraud detection method and device, this method include：Obtain the authorization data sent after target user is authorized by internet credit APP；The feature vector of target user is built based on authorization data；Feature vector is trained by K MEANS algorithms, obtains unsupervised anti-fraud machine learning model；Probability of cheating calculating is carried out to feature vector by unsupervised anti-fraud machine learning model, obtains the probability of cheating of target user.This method takes full advantage of the authorization data of target user, feature vector is obtained to authorization data vectorization, and then it models to obtain unsupervised anti-fraud machine learning model by K MEANS algorithms, unsupervised anti-fraud machine learning model carries out probability of cheating calculating to feature vector again, obtain the probability of cheating of target user, this method can find new-type fraud pattern in real time, alleviate the technical issues of existing fraud detection method can not identify new-type fraud pattern within a short period of time.
Description
Technical field
The present invention relates to the technical fields of internet credit air control, more particularly, to the reality under a kind of internet credit scene
When fraud detection method and device.
Background technology
Internet credit industry, was developed rapidly in recent years, show let a hundred schools contend, the situation that a hundred flowers blossom, companion
With the glad flourish development of industry, fraud Dark Industry Link is also constantly penetrating into the field, various novel fraud mode layers
Go out not poor, one layer of shade has been coverd with to the sound development of internet credit industry.According to incompletely statistics, every year caused by fraud
500 hundred million -1,000 hundred million, risk of fraud has become the most important thing of internet credit industry risk for loss.
The main method of credit industry fraud prevention be rule-based engine method and be based on supervision machine learning model
Method, the method for rule-based engine is to take precautions against rule by converting the Heuristics of air control expert to fraud, passes through rule
Then the mode of engine is matched.Based on the method for supervision machine learning model, by that will have the crowd of fraud and not have
The crowd of fraud is combined as sample data, by choosing corresponding feature, using supervision machine learning method, builds mould
Type, to identify risk of fraud.
For above two method in traditional credit industry, effect is more apparent, however under internet credit scene,
Under the overall background of internet, innovative service is quickly grown, therefore, under various businesses scene fraudulent mean and technology also exist
It constantly updates, the method for rule-based engine and supervision machine learning model is all the mould obtained according to existing fraud pattern
Type can only identify existing fraud pattern, can not identify new-type fraud pattern within a short period of time.
To sum up, the method for existing credit industry fraud prevention can not identify new-type fraud mould within a short period of time
Formula.
Invention content
In view of this, the purpose of the present invention is to provide under a kind of internet credit scene real-time fraud detection method and
Device, the method to alleviate existing credit industry fraud prevention can not identify new-type fraud pattern within a short period of time
Technical problem.
In a first aspect, an embodiment of the present invention provides the real-time fraud detection method under a kind of internet credit scene, institute
The method of stating includes：
Obtain the authorization data sent after target user is authorized by internet credit APP, wherein the authorization data packets
It includes：The device data of the target user, the behavioral data of the target user, the social data of the target user are described
The application business datum of target user；
The feature vector of the target user is built based on the authorization data, wherein described eigenvector includes：Statistics
Feature vector, relationship characteristic vector, behavioural characteristic vector；
Described eigenvector is trained by K-MEANS algorithms, obtains unsupervised anti-fraud machine learning model；
Probability of cheating calculating is carried out to described eigenvector by the unsupervised anti-fraud machine learning model, obtains institute
State the probability of cheating of target user.
With reference to first aspect, an embodiment of the present invention provides the first possible embodiments of first aspect, wherein base
The feature vector that the target user is built in the authorization data includes：
The device data of behavioral data and the target user to the target user calculates, and obtains the statistics
Feature vector；
To the social data of the target user, the number of devices of the request for data of the target user and the target user
According to being calculated, the relationship characteristic vector is obtained；
The behavioral data of the target user is calculated, obtains the behavioural characteristic vector, wherein the behavior is special
Levying vector includes：Input behavior feature vector, operation behavior feature vector.
With reference to first aspect, an embodiment of the present invention provides second of possible embodiments of first aspect, wherein right
The behavioral data of the target user and the device data of the target user calculate, and obtain the statistical nature vector packet
It includes：
Obtain the target signature range of the target signature range and fraud crowd of non-fraud crowd；
The device data of behavioral data and the target user based on the target user extracts the target user's
Target signature, wherein the target signature of the target user includes：The geography information application frequency, the application frequency of IP, equipment electricity
Measure accounting, the average acceleration of gyroscope；
In conjunction with the target signature range of the non-fraud crowd, the target signature range of the fraud crowd and the mesh
The target signature for marking user, determines the Crowds Distribute belonging to the target user；
The statistical nature vector is calculated based on the Crowds Distribute belonging to the target user.
With reference to first aspect, an embodiment of the present invention provides the third possible embodiments of first aspect, wherein right
The device data of the social data of the target user, the request for data of the target user and the target user are counted
It calculates, obtaining the relationship characteristic vector includes：
By the device data of the target user, the application of the social data of the target user and the target user
Data are associated with the foundation of historical relation collection of illustrative plates, wherein the historical relation collection of illustrative plates is the relationship obtained according to history authorization data
Collection of illustrative plates；
The historical relation collection of illustrative plates is calculated by community discovery algorithm, obtains the social activity belonging to the target user
Group；
The weighted value that side in the historical relation collection of illustrative plates is updated by the risk of fraud of the social groups, after obtaining update
Relation map, wherein the risk of fraud of the social groups is to be obtained according to the history authorization data；
The updated relation map is calculated by Random Walk Algorithm and node2vector, is obtained described
Relationship characteristic vector.
With reference to first aspect, an embodiment of the present invention provides the 4th kind of possible embodiments of first aspect, wherein right
The behavioral data of the target user calculates, and obtains the behavioural characteristic vector and includes：
Input behavior data are extracted in the behavioral data of the target user；
The input total time-consuming of input behavior is calculated according to the input behavior data, is inputted and is averagely taken, inputs character
Equispaced takes, variance, wherein the input behavior includes：The behavior of input identification card number, the behavior of input handset number,
The behavior of bank's card number is inputted, the variance is used to indicate the fluctuation situation of input character pitch；
By the input total time-consuming, the input is average to be taken, and the equispaced of the input character takes, the variance
The input behavior feature vector as the target user；
The extraction operation behavioral data in the behavioral data of the target user；
The operation behavior data are analyzed, the operation time-delay series of operation behavior are obtained, wherein the operation row
For the behavior for the operation internet credit APP；
Using the operation time-delay series as the operation behavior feature vector.
With reference to first aspect, an embodiment of the present invention provides the 5th kind of possible embodiments of first aspect, wherein institute
The method of stating further includes：
Probability of cheating is carried out to the group belonging to the target user by the unsupervised anti-fraud machine learning model
It calculates, obtains the probability of cheating of the group.
With reference to first aspect, an embodiment of the present invention provides the 6th kind of possible embodiments of first aspect, wherein institute
The method of stating further includes：
The operation behavior of the group belonging to the target user is calculated by statistical analysis technique, obtains the group
The behavior pattern vector of body；
The behavior pattern vector is monitored in real time；
When significant changes occurs in the behavior pattern vector, determine that the group is with preclinical fraud group.
Second aspect, the embodiment of the present invention additionally provide the real-time fraud detection device under a kind of internet credit scene,
Described device includes：
Acquisition module, for obtaining the authorization data sent after target user is authorized by internet credit APP, wherein
The authorization data includes：The device data of the target user, the behavioral data of the target user, the target user's
Social data, the application business datum of the target user；
Build module, the feature vector for building the target user based on the authorization data, wherein the feature
Vector includes：Statistical nature vector, relationship characteristic vector, behavioural characteristic vector；
Training module is trained described eigenvector for passing through K-MEANS algorithms, obtains unsupervised anti-fraud machine
Device learning model；
First probability of cheating computing module, for by the unsupervised anti-fraud machine learning model to the feature to
Amount carries out probability of cheating calculating, obtains the probability of cheating of the target user.
In conjunction with second aspect, an embodiment of the present invention provides the first possible embodiments of second aspect, wherein institute
Stating structure module includes：
First computing unit, the device data for behavioral data and the target user to the target user carry out
It calculates, obtains the statistical nature vector；
Second computing unit, for the social data to the target user, the request for data of the target user and institute
The device data for stating target user calculates, and obtains the relationship characteristic vector；
Third computing unit is calculated for the behavioral data to the target user, obtain the behavioural characteristic to
Amount, wherein the behavioural characteristic vector includes：Input behavior feature vector, operation behavior feature vector.
In conjunction with second aspect, an embodiment of the present invention provides second of possible embodiments of second aspect, wherein institute
Stating the first computing unit includes：
Subelement is obtained, the target signature range of target signature range and fraud crowd for obtaining non-fraud crowd；
First extraction subelement, is used for the device data of behavioral data and the target user based on the target user
Extract the target signature of the target user, wherein the target signature of the target user includes：The geography information application frequency,
The application frequency of IP, equipment electricity accounting, the average acceleration of gyroscope；
Determination subelement, for the target signature range in conjunction with the non-fraud crowd, the target of the fraud crowd is special
Range and the target signature of the target user are levied, determines the Crowds Distribute belonging to the target user；
First computation subunit, for based on the Crowds Distribute belonging to the target user calculate the statistical nature to
Amount.
The embodiment of the present invention brings following advantageous effect：An embodiment of the present invention provides under a kind of internet credit scene
Real-time fraud detection method and device, this method include：What acquisition target user sent after being authorized by internet credit APP
Authorization data, wherein authorization data includes：The device data of target user, the behavioral data of target user, the society of target user
Intersection number evidence, the application business datum of target user；Based on authorization data build target user feature vector, wherein feature to
Amount includes：Statistical nature vector, relationship characteristic vector, behavioural characteristic vector；Feature vector is instructed by K-MEANS algorithms
Practice, obtains unsupervised anti-fraud machine learning model；Feature vector is taken advantage of by unsupervised anti-fraud machine learning model
Probability calculation is cheated, the probability of cheating of target user is obtained.
The fraud detection method of existing rule-based engine and supervision machine learning model is all according to existing fraud
The model that pattern obtains can only identify existing fraud pattern, can not identify new-type fraud pattern within a short period of time.With it is existing
The rule-based engine having is compared with the fraud detection method of supervision machine learning model, the internet letter in the embodiment of the present invention
It borrows in the real-time fraud detection method under scene, the authorization data vectorization of target user can be obtained the spy of target user
Sign vector, is trained feature vector by K-MEANS algorithms, obtains unsupervised anti-fraud machine learning model, finally, leads to
It crosses unsupervised anti-fraud machine learning model and probability of cheating calculating is carried out to the feature vector of target user, just can obtain target
The probability of cheating of user.This method takes full advantage of the authorization data of target user, if target user is with the mesh of premeditated fraud
Carry out credit applications, then the clues and traces cheated can be hidden in authorization data, and authorization data is unstructured data, can not
It is modeled for K-MEANS algorithms, so needing to carry out vectorization, obtains the feature vector of target user, and then model
To unsupervised anti-fraud machine learning model, finally, feature vector is taken advantage of by unsupervised anti-fraud machine learning model
Probability calculation is cheated, the probability of cheating of target user is obtained, this method can find new-type fraud pattern in real time, help internet
Credit industry successfully manages risk of fraud, alleviate existing fraud detection method can not identify within a short period of time it is new-type
The technical issues of fraud pattern.
Other features and advantages of the present invention will illustrate in the following description, also, partly become from specification
It obtains it is clear that understand through the implementation of the invention.The purpose of the present invention and other advantages are in specification, claims
And specifically noted structure is realized and is obtained in attached drawing.
To enable the above objects, features and advantages of the present invention to be clearer and more comprehensible, preferred embodiment cited below particularly, and coordinate
Appended attached drawing, is described in detail below.
Description of the drawings
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 are briefly described, it should be apparent that, in being 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, other drawings may also be obtained based on these drawings.
Fig. 1 is the flow of the real-time fraud detection method under a kind of internet credit scene provided in an embodiment of the present invention
Figure；
Fig. 2 is the flow chart of the feature vector provided in an embodiment of the present invention that target user is built based on authorization data；
Fig. 3 counts the behavioral data of target user and the device data of target user to be provided in an embodiment of the present invention
It calculates, obtains the flow chart of statistical nature vector；
Fig. 4 is the social data provided in an embodiment of the present invention to target user, the request for data and target of target user
The device data of user calculates, and obtains the flow chart of relationship characteristic vector；
Fig. 5 is that the behavioral data provided in an embodiment of the present invention to target user calculates, and obtains behavioural characteristic vector
Flow chart；
Fig. 6 is the signal of the real-time fraud detection device under a kind of internet credit scene provided in an embodiment of the present invention
Figure.
Icon：
11- acquisition modules；12- builds module；13- training modules；14- the first probability of cheating computing modules.
Specific implementation mode
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
Lower obtained every other embodiment, shall fall within the protection scope of the present invention.
For ease of understanding the present embodiment, first to a kind of internet credit scene disclosed in the embodiment of the present invention
Under real-time fraud detection method describe in detail.
Embodiment one：
A kind of real-time fraud detection method under internet credit scene, with reference to figure 1, this method includes：
S102, the authorization data sent after target user is authorized by internet credit APP is obtained, wherein authorization data
Including：The device data of target user, the behavioral data of target user, the social data of target user, the application of target user
Business datum；
In embodiments of the present invention, user generally requires when carrying out credit applications and uses internet credit in mobile phone terminal
APP fills in related data, and after filling in, mandate is applied, so, referred to as authorization data.
The executive agent of real-time fraud detection method under the internet credit scene is server, and server can be obtained and be awarded
Flexible strategy evidence.Specifically, authorization data includes：The device data of target user, the behavioral data of target user, the society of target user
Intersection number evidence, the application business datum of target user.
Device data includes：Mobile phone number, GPS location data, MAC Address data, IP address data etc., the present invention
Embodiment is not particularly limited it.
Behavioral data includes：Operation behavior data, input behavior data.Operation behavior data refer to operation internet credit
Data when APP；Input behavior data refer to input identification card number, and input handset number, input bank card number etc., the present invention is implemented
Example is not particularly limited it.
S104, the feature vector that target user is built based on authorization data, wherein feature vector includes：Statistical nature to
Amount, relationship characteristic vector, behavioural characteristic vector；
After authorized data, due to device data, behavioral data, social data is unstructured data, Wu Fayong
In the computation in later stage, need these unstructured datas carrying out structuring.Specifically, building target based on authorization data
The feature vector of user includes statistical nature vector, relationship characteristic vector, behavioural characteristic vector.
S106, feature vector is trained by K-MEANS algorithms, obtains unsupervised anti-fraud machine learning model；
After obtaining the feature vector of target user, feature vector is trained by K-MEANS algorithms, obtains no prison
Superintend and direct anti-fraud machine learning model.K-MEANS algorithms are one kind in unsupervised machine learning algorithm, and K- is being used in the present invention
When MEANS algorithms, K therein is obtained after being analyzed authorization data by Gaussian function.
S108, probability of cheating calculating is carried out to feature vector by unsupervised anti-fraud machine learning model, obtains target
The probability of cheating of user.
After obtaining unsupervised anti-fraud machine learning model, by unsupervised anti-fraud machine learning model to feature to
Amount carries out probability of cheating calculating, it will be able to obtain the probability of cheating of target user.
Specifically, in background server, unsupervised anti-fraud machine learning model has been constructed in advance.This is unsupervised
Anti- fraud machine learning model is to be built in the way of step S102 to step S106 according to a large amount of history authorization data
It arrives, after the completion of foundation, which carries out online real-time learning, and new authorization data carries out step S102 to step S108's
It calculates, obtains probability of cheating, and continuous iteration optimization is carried out to the model.
The fraud detection method of existing rule-based engine and supervision machine learning model is all according to existing fraud
The model that pattern obtains can only identify existing fraud pattern, can not identify new-type fraud pattern within a short period of time.With it is existing
The rule-based engine having is compared with the fraud detection method of supervision machine learning model, the internet letter in the embodiment of the present invention
It borrows in the real-time fraud detection method under scene, the authorization data vectorization of target user can be obtained the spy of target user
Sign vector, is trained feature vector by K-MEANS algorithms, obtains unsupervised anti-fraud machine learning model, finally, leads to
It crosses unsupervised anti-fraud machine learning model and probability of cheating calculating is carried out to the feature vector of target user, just can obtain target
The probability of cheating of user.This method takes full advantage of the authorization data of target user, if target user is with the mesh of premeditated fraud
Carry out credit applications, then the clues and traces cheated can be hidden in authorization data, and authorization data is unstructured data, can not
It is modeled for K-MEANS algorithms, so needing to carry out vectorization, obtains the feature vector of target user, and then model
To unsupervised anti-fraud machine learning model, finally, feature vector is taken advantage of by unsupervised anti-fraud machine learning model
Probability calculation is cheated, the probability of cheating of target user is obtained, this method can find new-type fraud pattern in real time, help internet
Credit industry successfully manages risk of fraud, alleviate existing fraud detection method can not identify within a short period of time it is new-type
The technical issues of fraud pattern.
The above briefly describes the real-time fraud detection method under internet credit scene, below to wherein
The particular content being related to is described in detail.
Optionally, with reference to figure 2, the feature vector that target user is built based on authorization data includes：
S201, the behavioral data of target user and the device data of target user are calculated, obtain statistical nature to
Amount；
Specifically, statistical nature vector is calculated according to the behavioral data of target user and the device data of target user
It arrives, hereinafter specific calculating process is described in detail again, details are not described herein.
S202, the social data to target user, the request for data of target user and the device data of target user carry out
It calculates, obtains relationship characteristic vector；
Specifically, relationship characteristic vector is the social data according to target user, the request for data and target of target user
What the device data of user was calculated, hereinafter specific calculating process is described in detail again, details are not described herein.
S203, the behavioral data of target user is calculated, obtains behavioural characteristic vector, wherein behavioural characteristic vector
Including：Input behavior feature vector, operation behavior feature vector.
Specifically, behavioural characteristic vector is calculated according to the behavioral data of target user, it is equally hereinafter right again
The process is described in detail, and details are not described herein.
Optionally, with reference to figure 3, the device data of behavioral data and target user to target user calculates, and obtains
Statistical nature vector includes：
S301, the target signature range for obtaining non-fraud crowd and the target signature range for cheating crowd；
In embodiments of the present invention, target signature includes：The geography information application frequency, the application frequency of IP, equipment electricity
Accounting, the average acceleration of gyroscope.There are target signature ranges by non-fraud crowd, and there is also target signature models by fraud crowd
It encloses, obtaining for the range can be obtained by expert or experience.For example it is non-fraud people that the geography information application frequency, which is less than 10 times,
Group, it is fraud crowd that the geography information application frequency, which is more than or equal to 10 times, which is scheme in order to better understand the present invention,
It should not be used as implementing the present invention limitation of power.
Target signature can be one or more of features described above, can also include other feature, the embodiment of the present invention
It is not particularly limited.
The meaning of the geographical information applications frequency is illustrated below：The same period is detected, such as at 3 minutes
Interior somewhere (GPS positioning obtains) has an area of credit applications number within 10 kilometers is how many.Because many people are to belong to centralized swindleness
It deceives, geography information may be the credit applications concentrated and carried out for some region, this just belongs to dangerous.
The application frequency of IP refers to a people while operating how many IP progress credit applications.
Equipment electricity accounting refers to the fluctuation situation of the cell phone apparatus electricity of user, if do not fluctuated, can be identified as
Malice manipulates.
The average acceleration of gyroscope is 0 or very little, illustrates the occupation mode for not meeting normal person, can be identified as cheating
Suspicion.
The target of the device data extraction target user of S302, the behavioral data based on target user and target user are special
Sign, wherein the target signature of target user includes：The geography information application frequency, the application frequency of IP, equipment electricity accounting, top
The average acceleration of spiral shell instrument；
S303, in conjunction with the target signature range of non-fraud crowd, cheat target signature range and the target user of crowd
Target signature, determine the Crowds Distribute belonging to target user；
Specifically, after obtaining the target signature of target user, judge which mesh is the target signature of target user fall at
It marks in characteristic range, determination obtains the crowd belonging to target user.
S304, based on the Crowds Distribute counting statistics feature vector belonging to target user.
After obtaining the Crowds Distribute belonging to target user, it will be able to obtain statistical nature vector.Specifically, target user
Target signature it is different in the position of target signature range, obtained numerical value is also different, just can obtain statistical nature vector.
Optionally, with reference to figure 4, to the social data of target user, the request for data of target user and setting for target user
Standby data are calculated, and obtaining relationship characteristic vector includes：
S401, the device data by target user, the social data of target user and the request for data of target user with
Historical relation collection of illustrative plates establishes association, wherein historical relation collection of illustrative plates is the relation map obtained according to history authorization data；
In embodiments of the present invention, there are history authorization datas in server, so, it is corresponding that there is also historical relation figures
Spectrum.In the device data for obtaining target user, after the social data of target user and the request for data of target user, by those
Data are associated with the foundation of historical relation collection of illustrative plates, are A there are a MAC Address for example, in historical relation collection of illustrative plates, target user's
MAC Address in device data is also A, just can establish the authorization data of target user and being associated with for historical relation collection of illustrative plates.
S402, historical relation collection of illustrative plates is calculated by community discovery algorithm, obtains the social group belonging to target user
Body；
Establish with after being associated with of historical relation collection of illustrative plates, by community discovery algorithm to the historical relation figure after establishing association
Spectrum is calculated, and the social groups belonging to target user are obtained.
S403, the weighted value that side in historical relation collection of illustrative plates is updated by the risk of fraud of social groups, obtain updated
Relation map, wherein the risk of fraud of social groups is to be obtained according to history authorization data；
The weighted value that side in historical relation collection of illustrative plates is updated by the risk of fraud of social groups, obtains updated relational graph
Spectrum.Wherein, known to the risk of fraud of social groups.
S404, updated relation map is calculated by Random Walk Algorithm and node2vector, obtains relationship
Feature vector.
After obtaining updated relation map, by Random Walk Algorithm and node2vector to updated relationship
Collection of illustrative plates is calculated, and relationship characteristic vector is obtained.
Optionally, with reference to figure 5, the behavioral data of target user is calculated, obtaining behavioural characteristic vector includes：
S501, input behavior data are extracted in the behavioral data of target user；
S502, the input total time-consuming that input behavior is calculated according to input behavior data, input and averagely take, input character
Equispaced takes, variance, wherein input behavior includes：Input the behavior of identification card number, the behavior of input handset number, input
The behavior of bank's card number, variance are used to indicate the fluctuation situation of input character pitch；
S503, total time-consuming will be inputted, and will input and averagely takes, the equispaced for inputting character takes, and variance is used as target
The input behavior feature vector at family；
S504, the extraction operation behavioral data in the behavioral data of target user；
S505, operation behavior data are analyzed, obtains the operation time-delay series of operation behavior, wherein operation behavior
To operate the behavior of internet credit APP；
Specifically, when operation internet credit APP, if operating procedure is fixed, there are certain time delay sequences between often walking
Row.
S506, time-delay series will be operated as operation behavior feature vector.
Optionally, this method further includes：
Probability of cheating calculating is carried out to the group belonging to target user by unsupervised anti-fraud machine learning model, is obtained
The probability of cheating of group.
The group belonging to target user is found by unsupervised anti-fraud machine learning model, by correlation analysis, is known
It not to be not whether fraud clique.
If saying that cluster obtains 100 groups, wherein there are one class is very big with other class difference, there are a classes to deviate
Other 99 classes, if it is very big with other group distance difference there are one group, illustrate that it is exactly abnormal point, probability of cheating is just
Greatly；
It is also to see whether individual with its other individual has similitude when cluster, has similitude to be just classified as one certainly
Inside a group, if he does not have similitude, illustrate him except group, he is exactly abnormal point, and probability of cheating is with regard to big.
Optionally, this method further includes：
(1) operation behavior of the group belonging to target user is calculated by statistical analysis technique, obtains group
Behavior pattern vector；
(2) behavior pattern vector is monitored in real time；
(3) when significant changes occurs in behavior pattern vector, determine that group is with preclinical fraud group.
The invention discloses the real-time fraud detection methods and device under a kind of internet credit scene, to unsupervised machine
The technological innovation of learning model is applied in the anti-fraud detection of air control of internet credit, passes through behavioral data to user and pass
The vectorization of coefficient evidence calculates, and unsupervised anti-fraud machine learning is built by K-MEANS algorithms in conjunction with other structures feature
Model can identify new-type fraud pattern and the fraud pattern of submarine in real time, improve that internet credit air control is counter to cheat
The timeliness and compliance of method effectively reduce the risk of fraud in internet credit scene.
Under actual internet credit scene, by the APP of mobile phone terminal, under the premise of user authorizes, user is obtained
Cell phone apparatus data, behavioral data, relation data cheat if the user carries out credit applications with the purpose of premeditated fraud
Clues and traces can be hidden in corresponding data.Due to non-structured behavior and relation data, engineering can not be applied to
Algorithm is practised to be modeled, by by relation data and behavioral data progress vectorization calculating, being converted to the data characteristics of structuring,
All features of user are finally combined as feature vector, by K-MEANS algorithms, calculate whether user belongs to credit applications use
Abnormal point in family is calculated by being associated with, judges whether user belongs to fraud gang member, it is anti-to improve internet credit air control
The timeliness and compliance of fraud method effectively reduce the risk of fraud in internet credit scene.
Embodiment two：
A kind of real-time fraud detection device under internet credit scene, with reference to figure 6, which includes：
Acquisition module 11, for obtaining the authorization data sent after target user is authorized by internet credit APP,
In, authorization data includes：The device data of target user, the behavioral data of target user, the social data of target user, target
The application business datum of user；
Module 12 is built, the feature vector for building target user based on authorization data, wherein feature vector includes：
Statistical nature vector, relationship characteristic vector, behavioural characteristic vector；
Training module 13 is trained feature vector for passing through K-MEANS algorithms, obtains unsupervised anti-fraud machine
Learning model；
First probability of cheating computing module 14, for being carried out to feature vector by unsupervised anti-fraud machine learning model
Probability of cheating calculates, and obtains the probability of cheating of target user.
It, can be by target user's in the real-time fraud detection device under internet credit scene in the embodiment of the present invention
Authorization data vectorization obtains the feature vector of target user, is trained, is obtained to feature vector by K-MEANS algorithms
Unsupervised anti-fraud machine learning model, finally, by unsupervised anti-fraud machine learning model to the feature of target user to
Amount carries out probability of cheating calculating, just can obtain the probability of cheating of target user.The device takes full advantage of awarding for target user
Flexible strategy evidence, if target user carries out credit applications with the purpose of premeditated fraud, the clues and traces cheated can be hidden in mandate
In data, authorization data is unstructured data, is not used to K-MEANS algorithms and is modeled, so need to carry out vectorization,
The feature vector of target user is obtained, and then models and obtains unsupervised anti-fraud machine learning model, finally, by unsupervised anti-
It cheats machine learning model and probability of cheating calculating is carried out to feature vector, obtain the probability of cheating of target user, which can
New-type fraud pattern is found in real time, internet credit industry is helped to successfully manage risk of fraud, alleviates existing fraud inspection
Survey the technical issues of device can not identify new-type fraud pattern within a short period of time.
Optionally, structure module includes：
First computing unit, the device data for behavioral data and target user to target user are calculated, are obtained
To statistical nature vector；
Second computing unit, for the social data of target user, the request for data of target user and target user's
Device data is calculated, and relationship characteristic vector is obtained；
Third computing unit is calculated for the behavioral data to target user, obtains behavioural characteristic vector, wherein
Behavioural characteristic vector includes：Input behavior feature vector, operation behavior feature vector.
Optionally, the first computing unit includes：
Subelement is obtained, the target signature range of target signature range and fraud crowd for obtaining non-fraud crowd；
First extraction subelement, the device data for behavioral data and target user based on target user extract target
The target signature of user, wherein the target signature of target user includes：The geography information application frequency, the application frequency of IP, equipment
Electricity accounting, the average acceleration of gyroscope；
Determination subelement, for combine non-fraud crowd target signature range, cheat the target signature range of crowd with
And the target signature of target user, determine the Crowds Distribute belonging to target user；
First computation subunit, for based on the Crowds Distribute counting statistics feature vector belonging to target user.
Optionally, the second computing unit includes：
Association subelement is established, for using the social data and target of the device data of target user, target user
The request for data at family is associated with the foundation of historical relation collection of illustrative plates, wherein historical relation collection of illustrative plates is obtained according to history authorization data
Relation map；
Second computation subunit calculates historical relation collection of illustrative plates for passing through community discovery algorithm, obtains target use
Social groups belonging to family；
Subelement is updated, the weighted value for updating side in historical relation collection of illustrative plates by the risk of fraud of social groups obtains
To updated relation map, wherein the risk of fraud of social groups is to be obtained according to history authorization data；
Third computation subunit, for by Random Walk Algorithm and node2vector to updated relation map into
Row calculates, and obtains relationship characteristic vector.
Optionally, third computing unit includes：
Second extraction subelement, for extracting input behavior data in the behavioral data of target user；
4th computation subunit, the input total time-consuming for calculating input behavior according to input behavior data, input are average
It takes, the equispaced for inputting character takes, variance, wherein input behavior includes：The behavior of identification card number is inputted, hand is inputted
The behavior of machine number, the behavior of input bank card number, variance are used to indicate the fluctuation situation of input character pitch；
First setting subelement, for that will input total time-consuming, input is average to be taken, and the equispaced for inputting character takes,
Input behavior feature vector of the variance as target user；
Third extracts subelement, for the extraction operation behavioral data in the behavioral data of target user；
Subelement is analyzed, for analyzing operation behavior data, obtains the operation time-delay series of operation behavior,
In, operation behavior is to operate the behavior of internet credit APP；
Second setting subelement, for time-delay series will to be operated as operation behavior feature vector.
Optionally, which further includes：
Second probability of cheating computing module is used for through unsupervised anti-fraud machine learning model to belonging to target user
Group carries out probability of cheating calculating, obtains the probability of cheating of group.
Optionally, which further includes：
Computing module, for being calculated the operation behavior of the group belonging to target user by statistical analysis technique,
Obtain the behavior pattern vector of group；
Real-time monitoring module, for being monitored in real time to behavior pattern vector；
Determining module, for when significant changes occurs in behavior pattern vector, determining that group is with preclinical fraud
Group.
Particular content in the embodiment two can refer to the specific descriptions in above-described embodiment one, and details are not described herein.
The computer of real-time fraud detection method and device under the internet credit scene that the embodiment of the present invention is provided
Program product, including the computer readable storage medium of program code is stored, the instruction that said program code includes can be used for
The method described in previous methods embodiment is executed, specific implementation can be found in embodiment of the method, and details are not described herein.
It is apparent to those skilled in the art that for convenience and simplicity of description, the system of foregoing description
It with the specific work process of device, can refer to corresponding processes in the foregoing method embodiment, details are not described herein.
In addition, in the description of the embodiment of the present invention unless specifically defined or limited otherwise, term " installation ", " phase
Even ", " connection " shall be understood in a broad sense, for example, it may be being fixedly connected, may be a detachable connection, or be integrally connected；It can
Can also be electrical connection to be mechanical connection；It can be directly connected, can also indirectly connected through an intermediary, Ke Yishi
Connection inside two elements.For the ordinary skill in the art, above-mentioned term can be understood at this with concrete condition
Concrete meaning in invention.
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 read/write memory medium.Based on this understanding, technical scheme 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 expressed 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
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：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 disc or CD.
In the description of the present invention, it should be noted that term "center", "upper", "lower", "left", "right", "vertical",
The orientation or positional relationship of the instructions such as "horizontal", "inner", "outside" be based on the orientation or positional relationship shown in the drawings, merely to
Convenient for the description present invention and simplify description, do not indicate or imply the indicated device or element must have a particular orientation,
With specific azimuth configuration and operation, therefore it is not considered as limiting the invention.In addition, term " first ", " second ",
" third " is used for description purposes only, and is not understood to indicate or imply relative importance.
Finally it should be noted that：Embodiment described above, only specific implementation mode 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, it will be understood by those of ordinary skill in the art that：Any one skilled in the art
In the technical scope disclosed by the present invention, it can still modify to the technical solution recorded in 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 the protection in the present 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. the real-time fraud detection method under a kind of internet credit scene, which is characterized in that the method includes：
Obtain the authorization data that sends after target user is authorized by internet credit APP, wherein the authorization data includes：
The device data of the target user, the behavioral data of the target user, the social data of the target user, the target
The application business datum of user；
The feature vector of the target user is built based on the authorization data, wherein described eigenvector includes：Statistical nature
Vector, relationship characteristic vector, behavioural characteristic vector；
Described eigenvector is trained by K-MEANS algorithms, obtains unsupervised anti-fraud machine learning model；
Probability of cheating calculating is carried out to described eigenvector by the unsupervised anti-fraud machine learning model, obtains the mesh
Mark the probability of cheating of user.
2. according to the method described in claim 1, it is characterized in that, building the spy of the target user based on the authorization data
Levying vector includes：
The device data of behavioral data and the target user to the target user calculates, and obtains the statistical nature
Vector；
To the social data of the target user, the device data of the request for data of the target user and the target user into
Row calculates, and obtains the relationship characteristic vector；
The behavioral data of the target user is calculated, behavioural characteristic vector is obtained, wherein the behavioural characteristic to
Amount includes：Input behavior feature vector, operation behavior feature vector.
3. according to the method described in claim 2, it is characterized in that, behavioral data and the target to the target user are used
The device data at family is calculated, and is obtained the statistical nature vector and is included：
Obtain the target signature range of the target signature range and fraud crowd of non-fraud crowd；
The device data of behavioral data and the target user based on the target user extracts the target of the target user
Feature, wherein the target signature of the target user includes：The geography information application frequency, the application frequency of IP, equipment electricity account for
Than the average acceleration of gyroscope；
In conjunction with the target signature range of the non-fraud crowd, the target signature range of the fraud crowd and the target are used
The target signature at family determines the Crowds Distribute belonging to the target user；
The statistical nature vector is calculated based on the Crowds Distribute belonging to the target user.
4. according to the method described in claim 2, it is characterized in that, social data to the target user, the target are used
The request for data at family and the device data of the target user calculate, and obtain the relationship characteristic vector and include：
By the device data of the target user, the request for data of the social data of the target user and the target user
It is associated with the foundation of historical relation collection of illustrative plates, wherein the historical relation collection of illustrative plates is the relation map obtained according to history authorization data；
The historical relation collection of illustrative plates is calculated by community discovery algorithm, obtains the social group belonging to the target user
Body；
The weighted value that side in the historical relation collection of illustrative plates is updated by the risk of fraud of the social groups, obtains updated pass
It is collection of illustrative plates, wherein the risk of fraud of the social groups is to be obtained according to the history authorization data；
The updated relation map is calculated by Random Walk Algorithm and node2vector, obtains the relationship
Feature vector.
5. according to the method described in claim 2, it is characterized in that, calculate the behavioral data of the target user, obtain
Include to the behavioural characteristic vector：
Input behavior data are extracted in the behavioral data of the target user；
The input total time-consuming of input behavior is calculated according to the input behavior data, is inputted and is averagely taken, inputs being averaged for character
Interval takes, variance, wherein the input behavior includes：Input the behavior of identification card number, the behavior of input handset number, input
The behavior of bank's card number, the variance are used to indicate the fluctuation situation of input character pitch；
By the input total time-consuming, the input is average to be taken, and the equispaced of the input character takes, the variance conduct
The input behavior feature vector of the target user；
The extraction operation behavioral data in the behavioral data of the target user；
The operation behavior data are analyzed, the operation time-delay series of operation behavior are obtained, wherein the operation behavior is
Operate the behavior of the internet credit APP；
Using the operation time-delay series as the operation behavior feature vector.
6. according to the method described in claim 1, it is characterized in that, the method further includes：
Probability of cheating calculating is carried out to the group belonging to the target user by the unsupervised anti-fraud machine learning model,
Obtain the probability of cheating of the group.
7. according to the method described in claim 1, it is characterized in that, the method further includes：
The operation behavior of the group belonging to the target user is calculated by statistical analysis technique, obtains the group
Behavior pattern vector；
The behavior pattern vector is monitored in real time；
When significant changes occurs in the behavior pattern vector, determine that the group is with preclinical fraud group.
8. the real-time fraud detection device under a kind of internet credit scene, which is characterized in that described device includes：
Acquisition module, for obtaining the authorization data sent after target user is authorized by internet credit APP, wherein described
Authorization data includes：The device data of the target user, the behavioral data of the target user, the social activity of the target user
Data, the application business datum of the target user；
Build module, the feature vector for building the target user based on the authorization data, wherein described eigenvector
Including：Statistical nature vector, relationship characteristic vector, behavioural characteristic vector；
Training module is trained described eigenvector for passing through K-MEANS algorithms, obtains unsupervised anti-fraud engineering
Practise model；
First probability of cheating computing module, for by the unsupervised anti-fraud machine learning model to described eigenvector into
Row probability of cheating calculates, and obtains the probability of cheating of the target user.
9. device according to claim 8, which is characterized in that the structure module includes：
First computing unit, carry out by the device data of behavioral data and the target user to the target user based on
It calculates, obtains the statistical nature vector；
Second computing unit, for the social data to the target user, the request for data of the target user and the mesh
The device data of mark user calculates, and obtains the relationship characteristic vector；
Third computing unit is calculated for the behavioral data to the target user, obtains the behavioural characteristic vector,
In, the behavioural characteristic vector includes：Input behavior feature vector, operation behavior feature vector.
10. device according to claim 9, which is characterized in that first computing unit includes：
Subelement is obtained, the target signature range of target signature range and fraud crowd for obtaining non-fraud crowd；
First extraction subelement, the device data for behavioral data and the target user based on the target user extract
The target signature of the target user, wherein the target signature of the target user includes：The geography information application frequency, IP's
Apply for the frequency, equipment electricity accounting, the average acceleration of gyroscope；
Determination subelement, for the target signature range in conjunction with the non-fraud crowd, the target signature model of the fraud crowd
It encloses and the target signature of the target user, determines the Crowds Distribute belonging to the target user；
First computation subunit, for calculating the statistical nature vector based on the Crowds Distribute belonging to the target user.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810033615.2A CN108564460B (en) | 2018-01-12 | 2018-01-12 | Real-time fraud detection method and device in internet credit scene |
| PCT/CN2018/109729 WO2019137050A1 (en) | 2018-01-12 | 2018-10-10 | Real-time fraud detection method and device under internet credit scene, and server |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810033615.2A CN108564460B (en) | 2018-01-12 | 2018-01-12 | Real-time fraud detection method and device in internet credit scene |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108564460A true CN108564460A (en) | 2018-09-21 |
| CN108564460B CN108564460B (en) | 2020-10-30 |
Family
ID=63529793
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810033615.2A Active CN108564460B (en) | 2018-01-12 | 2018-01-12 | Real-time fraud detection method and device in internet credit scene |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN108564460B (en) |
| WO (1) | WO2019137050A1 (en) |
Cited By (20)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109583782A (en) * | 2018-12-07 | 2019-04-05 | 厦门铅笔头信息科技有限公司 | Support the auto metal halide lamp air control model of multi-data source |
| CN109685647A (en) * | 2018-12-27 | 2019-04-26 | 阳光财产保险股份有限公司 | The training method of credit fraud detection method and its model, device and server |
| CN109754258A (en) * | 2018-12-24 | 2019-05-14 | 同济大学 | It is a kind of based on individual behavior modeling towards online trading fraud detection method |
| CN109840778A (en) * | 2018-12-21 | 2019-06-04 | 上海拍拍贷金融信息服务有限公司 | The recognition methods of fraudulent user and device, readable storage medium storing program for executing |
| CN109992578A (en) * | 2019-01-07 | 2019-07-09 | 平安科技（深圳）有限公司 | Anti- fraud method, apparatus, computer equipment and storage medium based on unsupervised learning |
| CN110009174A (en) * | 2018-12-13 | 2019-07-12 | 阿里巴巴集团控股有限公司 | Risk identification model training method, device and server |
| WO2019137050A1 (en) * | 2018-01-12 | 2019-07-18 | 阳光财产保险股份有限公司 | Real-time fraud detection method and device under internet credit scene, and server |
| CN110148001A (en) * | 2019-04-29 | 2019-08-20 | 上海欣方智能系统有限公司 | A kind of system and method for realizing fraudulent trading intelligent early-warning |
| CN110363406A (en) * | 2019-06-27 | 2019-10-22 | 上海淇馥信息技术有限公司 | Appraisal procedure, device and the electronic equipment of a kind of client intermediary risk |
| 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 |
| CN111127026A (en) * | 2019-12-13 | 2020-05-08 | 深圳中兴飞贷金融科技有限公司 | Method, device, storage medium and electronic equipment for determining user fraud behavior |
| 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 |
| 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 |
| CN111385420A (en) * | 2018-12-29 | 2020-07-07 | 中兴通讯股份有限公司 | User identification method and device |
| CN111476653A (en) * | 2019-12-24 | 2020-07-31 | 马上消费金融股份有限公司 | Risk information identification, determination and model training method and device |
| CN111639681A (en) * | 2020-05-09 | 2020-09-08 | 同济大学 | Early warning method, system, medium and device based on education drive type fraud |
| CN111932269A (en) * | 2020-08-11 | 2020-11-13 | 中国工商银行股份有限公司 | Equipment information processing method and device |
| WO2020253358A1 (en) * | 2019-06-18 | 2020-12-24 | 深圳壹账通智能科技有限公司 | Service data risk control analysis processing method, apparatus and computer device |
| CN112819611A (en) * | 2021-03-02 | 2021-05-18 | 成都新希望金融信息有限公司 | Fraud identification method, device, electronic equipment and computer-readable storage medium |
| CN113706176A (en) * | 2021-09-02 | 2021-11-26 | 赵琦 | Information anti-fraud processing method and service platform system combined with cloud computing |
Families Citing this family (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110852761B (en) * | 2019-10-11 | 2023-07-04 | 支付宝(杭州)信息技术有限公司 | Method and device for formulating anti-cheating strategy and electronic equipment |
| CN112860949A (en) * | 2019-11-27 | 2021-05-28 | 国网电子商务有限公司 | Method and device for extracting map features |
| CN111222976B (en) * | 2019-12-16 | 2024-04-23 | 北京淇瑀信息科技有限公司 | Risk prediction method and device based on network map data of two parties and electronic equipment |
| CN111309822B (en) * | 2020-02-11 | 2023-05-09 | 简链科技(广东)有限公司 | User identity recognition method and device |
| CN112241549A (en) * | 2020-05-26 | 2021-01-19 | 中国银联股份有限公司 | Secure privacy calculation method, server, system, and storage medium |
| CN111881991B (en) * | 2020-08-03 | 2023-11-10 | 联仁健康医疗大数据科技股份有限公司 | Method and device for identifying fraud and electronic equipment |
| CN112200583B (en) * | 2020-10-28 | 2023-12-19 | 交通银行股份有限公司 | Knowledge graph-based fraudulent client identification method |
| US11288668B1 (en) | 2021-02-16 | 2022-03-29 | Capital One Services, Llc | Enhanced feedback exposure for users based on transaction metadata |
| US11443312B2 (en) | 2021-02-16 | 2022-09-13 | Capital One Services, Llc | Enhanced feedback exposure for merchants based on transaction metadata |
| US11257083B1 (en) | 2021-02-16 | 2022-02-22 | Capital One Services, Llc | Dynamic transaction metadata validation adjustment based on network conditions |
| US11182797B1 (en) | 2021-02-16 | 2021-11-23 | Capital One Services, Llc | Direct data share |
| CN113094506B (en) * | 2021-04-14 | 2023-08-18 | 每日互动股份有限公司 | Early warning method based on relational graph, computer equipment and 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 |
| CN113850665B (en) * | 2021-09-14 | 2023-09-12 | 江苏中交车旺科技有限公司 | Method and system for preventing and controlling fraud based on logistic finance knowledge graph |
| CN115730831B (en) * | 2023-01-10 | 2023-06-06 | 北京迈道科技有限公司 | Safety index evaluation method and device for construction work organization behaviors and electronic equipment |
| CN115860751A (en) * | 2023-02-27 | 2023-03-28 | 天津金城银行股份有限公司 | Anti-fraud analysis processing method and device and electronic equipment |
| CN116542673B (en) * | 2023-07-05 | 2023-09-08 | 成都乐超人科技有限公司 | Fraud identification method and system applied to machine learning |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103279868A (en) * | 2013-05-22 | 2013-09-04 | 兰亭集势有限公司 | Method and device for automatically identifying fraud order form |
| CN105516152A (en) * | 2015-12-15 | 2016-04-20 | 云南大学 | Abnormal behavior detection method |
| CN105512938A (en) * | 2016-02-03 | 2016-04-20 | 宜人恒业科技发展（北京）有限公司 | Online credit risk assessment method based on long-term using behavior of user |
| CN105894372A (en) * | 2016-06-13 | 2016-08-24 | 腾讯科技（深圳）有限公司 | Method and device for predicting group credit |
| CN106682985A (en) * | 2016-12-26 | 2017-05-17 | 深圳先进技术研究院 | Financial fraud identification method and system thereof |
| US20170364933A1 (en) * | 2014-12-09 | 2017-12-21 | Beijing Didi Infinity Technology And Development Co., Ltd. | User maintenance system and method |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106611375A (en) * | 2015-10-22 | 2017-05-03 | 北京大学 | Text analysis-based credit risk assessment method and apparatus |
| CN106529773A (en) * | 2016-10-31 | 2017-03-22 | 宜人恒业科技发展（北京）有限公司 | Online credit and fraud risk evaluation method based on identifying code type question answering |
| CN107194803A (en) * | 2017-05-19 | 2017-09-22 | 南京工业大学 | A kind of P2P nets borrow the device of borrower's assessing credit risks |
| CN108564460B (en) * | 2018-01-12 | 2020-10-30 | 阳光财产保险股份有限公司 | Real-time fraud detection method and device in internet credit scene |
-
2018
- 2018-01-12 CN CN201810033615.2A patent/CN108564460B/en active Active
- 2018-10-10 WO PCT/CN2018/109729 patent/WO2019137050A1/en active Application Filing
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103279868A (en) * | 2013-05-22 | 2013-09-04 | 兰亭集势有限公司 | Method and device for automatically identifying fraud order form |
| US20170364933A1 (en) * | 2014-12-09 | 2017-12-21 | Beijing Didi Infinity Technology And Development Co., Ltd. | User maintenance system and method |
| CN105516152A (en) * | 2015-12-15 | 2016-04-20 | 云南大学 | Abnormal behavior detection method |
| CN105512938A (en) * | 2016-02-03 | 2016-04-20 | 宜人恒业科技发展（北京）有限公司 | Online credit risk assessment method based on long-term using behavior of user |
| CN105894372A (en) * | 2016-06-13 | 2016-08-24 | 腾讯科技（深圳）有限公司 | Method and device for predicting group credit |
| CN106682985A (en) * | 2016-12-26 | 2017-05-17 | 深圳先进技术研究院 | Financial fraud identification method and system thereof |
Cited By (30)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2019137050A1 (en) * | 2018-01-12 | 2019-07-18 | 阳光财产保险股份有限公司 | Real-time fraud detection method and device under internet credit scene, and server |
| CN109583782A (en) * | 2018-12-07 | 2019-04-05 | 厦门铅笔头信息科技有限公司 | Support the auto metal halide lamp air control model of multi-data source |
| WO2020119272A1 (en) * | 2018-12-13 | 2020-06-18 | 阿里巴巴集团控股有限公司 | Risk identification model training method and apparatus, and server |
| TWI712981B (en) * | 2018-12-13 | 2020-12-11 | 開曼群島商創新先進技術有限公司 | Risk identification model training method, device and server |
| CN110009174B (en) * | 2018-12-13 | 2020-11-06 | 创新先进技术有限公司 | Risk recognition model training method and device and server |
| CN110009174A (en) * | 2018-12-13 | 2019-07-12 | 阿里巴巴集团控股有限公司 | Risk identification model training method, device and server |
| CN109840778A (en) * | 2018-12-21 | 2019-06-04 | 上海拍拍贷金融信息服务有限公司 | The recognition methods of fraudulent user and device, readable storage medium storing program for executing |
| CN109754258B (en) * | 2018-12-24 | 2023-05-12 | 同济大学 | Online transaction fraud detection method based on individual behavior modeling |
| CN109754258A (en) * | 2018-12-24 | 2019-05-14 | 同济大学 | It is a kind of based on individual behavior modeling towards online trading fraud detection method |
| CN109685647B (en) * | 2018-12-27 | 2021-08-10 | 阳光财产保险股份有限公司 | Credit fraud detection method and training method and device of model thereof, and server |
| CN109685647A (en) * | 2018-12-27 | 2019-04-26 | 阳光财产保险股份有限公司 | The training method of credit fraud detection method and its model, device and server |
| CN111385420B (en) * | 2018-12-29 | 2022-04-29 | 中兴通讯股份有限公司 | User identification method and device, storage medium and electronic device |
| CN111385420A (en) * | 2018-12-29 | 2020-07-07 | 中兴通讯股份有限公司 | User identification method and device |
| CN109992578A (en) * | 2019-01-07 | 2019-07-09 | 平安科技（深圳）有限公司 | Anti- fraud method, apparatus, computer equipment and storage medium based on unsupervised learning |
| CN109992578B (en) * | 2019-01-07 | 2023-08-08 | 平安科技（深圳）有限公司 | Anti-fraud method and device based on unsupervised learning, computer equipment and storage medium |
| CN110148001A (en) * | 2019-04-29 | 2019-08-20 | 上海欣方智能系统有限公司 | A kind of system and method for realizing fraudulent trading intelligent early-warning |
| WO2020253358A1 (en) * | 2019-06-18 | 2020-12-24 | 深圳壹账通智能科技有限公司 | Service data risk control analysis processing method, apparatus and computer device |
| CN110363406A (en) * | 2019-06-27 | 2019-10-22 | 上海淇馥信息技术有限公司 | Appraisal procedure, device and the electronic equipment of a kind of client intermediary risk |
| 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 |
| 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 |
| 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 |
| CN111127026A (en) * | 2019-12-13 | 2020-05-08 | 深圳中兴飞贷金融科技有限公司 | Method, device, storage medium and electronic equipment for determining user fraud behavior |
| CN111476653A (en) * | 2019-12-24 | 2020-07-31 | 马上消费金融股份有限公司 | Risk information identification, determination and model training method and device |
| 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 |
| CN111639681A (en) * | 2020-05-09 | 2020-09-08 | 同济大学 | Early warning method, system, medium and device based on education drive type fraud |
| CN111932269B (en) * | 2020-08-11 | 2023-08-18 | 中国工商银行股份有限公司 | Equipment information processing method and device |
| CN111932269A (en) * | 2020-08-11 | 2020-11-13 | 中国工商银行股份有限公司 | Equipment information processing method and device |
| CN112819611A (en) * | 2021-03-02 | 2021-05-18 | 成都新希望金融信息有限公司 | Fraud identification method, device, electronic equipment and computer-readable storage medium |
| CN113706176A (en) * | 2021-09-02 | 2021-11-26 | 赵琦 | Information anti-fraud processing method and service platform system combined with cloud computing |
| CN113706176B (en) * | 2021-09-02 | 2022-08-19 | 江西裕民银行股份有限公司 | Information anti-fraud processing method and service platform system combined with cloud computing |
Also Published As
| Publication number | Publication date |
|---|---|
| WO2019137050A1 (en) | 2019-07-18 |
| CN108564460B (en) | 2020-10-30 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108564460A (en) | Real-time fraud detection method under internet credit scene and device | |
| CN109948669A (en) | A kind of abnormal deviation data examination method and device | |
| CN109685647A (en) | The training method of credit fraud detection method and its model, device and server | |
| CN106503873A (en) | A kind of prediction user follows treaty method, device and the computing device of probability | |
| CN106953854B (en) | A kind of method for building up of the darknet flow identification model based on SVM machine learning | |
| CN108256431A (en) | A kind of hand position identification method and device | |
| CN107423742A (en) | The determination method and device of crowd's flow | |
| CN108121961A (en) | Inspection Activity recognition method, apparatus, computer equipment and storage medium | |
| CN110378699A (en) | A kind of anti-fraud method, apparatus and system of transaction | |
| CN109063977A (en) | A kind of no-induction transaction risk monitoring method and device | |
| Liu et al. | Focusing matching localization method based on indoor magnetic map | |
| CN109598374A (en) | A kind of heuristic efficiency analysis method of key facility physical protection system | |
| CN108960329A (en) | A kind of chemical process fault detection method comprising missing data | |
| CN107169769A (en) | The brush amount recognition methods of application program, device | |
| CN109508429A (en) | Personalized adaptive learning recommended method based on teaching platform big data analysis | |
| CN107273728B (en) | Smart watch unlocking and authentication method based on motion sensing behavior characteristics | |
| CN117495205B (en) | Industrial Internet experiment system and method | |
| CN110300385A (en) | A kind of indoor orientation method based on adaptive particle filter | |
| CN110263250A (en) | A kind of generation method and device of recommended models | |
| CN108712253A (en) | A kind of recognition methods of forgery mobile terminal and device based on mobile phone sensor fingerprint | |
| CN113256405A (en) | Method, device, equipment and storage medium for predicting cheating user concentrated area | |
| CN110909873B (en) | Method and device for denoising passive RFID (radio frequency identification) mobile ranging data and computer equipment | |
| CN110399279B (en) | Intelligent measurement method for non-human intelligent agent | |
| CN114299231A (en) | Construction method of 3D model for river water pollution | |
| TW201544830A (en) | Method of determining earthquake with artificial intelligence and earthquake detecting system |
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 |