Summary of the invention
The embodiment of the present application provides a kind of user type prediction technique, device, electronic equipment and storage medium, can be improved
Determine application user whether be black list user accuracy, convenient for improve risk identification ability.
In a first aspect, the embodiment of the present application provides a kind of user type prediction technique, comprising:
When receiving predictions request, user type prediction model is constructed;
Based on presetting database, the first user and the second user with first user-association are determined;
Based on the presetting database, the degree of association and described second for obtaining first user and the second user are used
The feature at family and its corresponding the First Eigenvalue;
The degree of association and the First Eigenvalue are input to the user type prediction model, described first is obtained and uses
The overdue probability at family;
Based on the overdue probability, predict that first user is black list user.
In one implementation, the building user type prediction model, comprising:
Based on the presetting database, determine that training application user collects corresponding training dataset and training result collection, tests
Card application user collects corresponding validation data set and verification result collection;
According to the training dataset and the training result collection, user type model of a syndrome to be tested is constructed;
According to the validation data set and the verification result collection, user type model of a syndrome to be tested is verified,
Obtain the user type prediction model.
In one implementation, described to be based on the presetting database, determine that training application user collects corresponding training
Data set and training result collection, verifying application user collect corresponding validation data set and verification result collection, comprising:
According to pre-set predetermined period, refund duration and the current time, determine the first period, the second period,
Third period and the 4th period;
In the presetting database, the Shen of first period, the second period authorized application information are obtained respectively
Please user, obtain the first user collection and second user collection;
In the presetting database, obtain at the beginning of from historical time to first period respectively, from
The application user of authorized application information and apply for user with it between at the beginning of the historical time to second period
Associated association user obtains third user collection and fourth user collection;
Determine that the user being overlapped between the first user collection and third user collection belongs to the training application user
The user being overlapped between collection, the second user collection and the fourth user collection belongs to verifying application user's collection;
In the presetting database, determined at the beginning of from the historical time to first period respectively
The training application user collection applies for user from the historical time to verifying described the end time of first period
Each application user or the corresponding feature of associated association user and its corresponding characteristic value are concentrated, the training data is obtained
Collection and the validation data set;
In the presetting database, determine respectively in the third period with the training application user collection, described the
The corresponding refund information of each apply user is concentrated with verifying application user in four periods, obtain the training dataset with
The verification result collection.
In one implementation, it is described be based on presetting database, determine the first user and with first user-association
Second user, comprising:
In the presetting database, the application for obtaining the authorized application information between historical time to current time is used
Family, obtains prediction application user's collection, and first user is any one application user that the prediction application user concentrates;
In the presetting database, the first user identifier based on first user search first user i.e.
When communication information and/or application information;
Based on the instant communication information and/or application information, the determining second user with first user-association.
In one implementation, described to be based on the presetting database, obtain the feature of the second user and its right
The First Eigenvalue answered, comprising:
In the presetting database, second user described in the second user identifier lookup based on the second user is corresponding
Multiple application informations;
If the number of the application information be it is multiple, obtain the corresponding objective appraisal value of each application information, obtain
Multiple objective appraisal values;
According to the multiple objective appraisal value obtain the second user feature and its corresponding the First Eigenvalue.
In one implementation, described to be based on the presetting database, obtain the feature of the second user and its right
The First Eigenvalue answered, comprising:
Based on the presetting database, the corresponding feature in the corresponding community of the second user and the community is determined
And its corresponding characteristic value;
Using the corresponding feature in the community and its corresponding characteristic value as the feature of the second user and its corresponding
The First Eigenvalue.
In one implementation, the quantity of the second user is multiple;
It is described that the degree of association and the First Eigenvalue are input to the user type prediction model, obtain described
The overdue probability of one user, comprising:
By the First Eigenvalue of each second user and the degree of association of first user, each second user
It is input to the user type prediction model, obtains the overdue probability of first user.
Second aspect, the embodiment of the present application provide a kind of user type prediction meanss, comprising:
Receiving unit, for receiving predictions request;
Processing unit, for constructing user type prediction model;Based on presetting database, determine the first user and with it is described
The second user of first user-association;Based on the presetting database, the pass of first user and the second user are obtained
The feature and its corresponding the First Eigenvalue of connection degree and the second user;The degree of association and the First Eigenvalue are inputted
To the user type prediction model, the overdue probability of first user is obtained;Based on the overdue probability, described the is predicted
One user is black list user.
In one implementation, in terms of the building user type prediction model, the processing unit is specifically used for
Determine that training application user collects corresponding training dataset and training result collection, verifying application user based on the presetting database
Collect corresponding validation data set and verification result collection;User type is constructed according to the training dataset and the training result collection
Model of a syndrome to be tested;User type model of a syndrome to be tested is verified according to the validation data set and the verification result collection
Obtain the user type prediction model.
In one implementation, determine that training application user collects corresponding training based on the presetting database described
In terms of data set and training result collection, verifying application user collect corresponding validation data set and verification result collection, the processing is single
Member be specifically used for according to pre-set predetermined period, refund duration and the current time determine the first period, the second period,
Third period and the 4th period;In the presetting database, first period, second period authorization Shen are obtained respectively
Please information application user obtain the first user collection and second user collection;In the presetting database, obtained respectively from history
Between at the beginning of time to first period, at the beginning of from the historical time to second period award
Weigh application information apply user and with its apply the association user of user-association obtain third user collect and fourth user collection;Really
The user being overlapped between the fixed first user collection and third user collection belongs to the training application user collection, described second
The user being overlapped between user's collection and the fourth user collection belongs to verifying application user's collection;In the presetting database
In, the training application user collection at the beginning of from the historical time to first period is determined respectively, from institute
It states historical time and concentrates each application user or its pass to verifying described between the end time of first period application user
The corresponding feature of the association user of connection and its corresponding characteristic value obtain the training dataset and the validation data set;Institute
State in presetting database, determine in the third period respectively with the training application user collection, in the 4th period with institute
It states verifying application user and concentrates the corresponding refund information of each application user, obtain the training dataset and the verification result
Collection.
In one implementation, it is described based on presetting database determine the first user and with first user-association
Second user in terms of, the processing unit be specifically used in the presetting database, obtain from historical time to it is current when
Between between authorized application information application user obtain prediction application user collection, first user be the prediction application user
Any one the application user concentrated;In the presetting database, the first user identifier based on first user is searched
The instant communication information and/or application information of first user;Based on the instant communication information and/or application information, really
The fixed second user with first user-association.
In one implementation, the feature of the second user and its right is obtained based on the presetting database described
In terms of the First Eigenvalue answered, the processing unit is specifically used in the presetting database, based on the second user
Second user described in second user identifier lookup corresponds to multiple application informations;If the number of the application information is multiple, acquisition
The corresponding objective appraisal value of each application information obtains multiple objective appraisal values;It is obtained according to the multiple objective appraisal value
The feature of the second user and its corresponding the First Eigenvalue.
In one implementation, the feature of the second user and its right is obtained based on the presetting database described
In terms of the First Eigenvalue answered, the processing unit is specifically used for determining that the second user is corresponding based on the presetting database
The corresponding feature in community and the community and its corresponding characteristic value;By the corresponding feature in the community and its corresponding spy
Feature and its corresponding the First Eigenvalue of the value indicative as the second user.
In one implementation, the quantity of the second user is multiple;Described by the degree of association and described
One characteristic value is input to the user type prediction model and obtains the overdue probability aspect of first user, the processing unit
Specifically for by the First Eigenvalue of the degree of association of each second user and first user, each second user
It is input to the user type prediction model and obtains the overdue probability of first user.
The third aspect, the embodiment of the present application provide a kind of electronic equipment, including processor, memory and one or more
Program, wherein said one or multiple programs are stored in above-mentioned memory, and are configured to be executed by above-mentioned processor,
Described program includes the instruction for the step some or all of as described in first aspect.
Fourth aspect, the embodiment of the present application provide a kind of computer readable storage medium, wherein described computer-readable
Storage medium stores computer program, wherein the computer program makes computer execute such as the embodiment of the present application first party
Step some or all of described in face.
5th aspect, the embodiment of the present application provide a kind of computer program product, wherein the computer program product
Non-transient computer readable storage medium including storing computer program, the computer program are operable to make to calculate
Machine executes the step some or all of as described in the embodiment of the present application first aspect.The computer program product can be one
A software installation packet.
By implementing the embodiment of the present invention, when receiving predictions request, user type prediction model is constructed, is then based on
Presetting database determines the first user and the second user with the first user-association, then obtains the first user based on presetting database
With the feature and its corresponding the First Eigenvalue of the degree of association of second user and second user, then the first user is used with second
The degree of association at family and the First Eigenvalue input user type prediction model obtain the overdue probability of the first user, based on overdue
The first user of probabilistic forecasting is black list user.I.e. the application can according to the characteristic value of other users (i.e. second user), with
The degree of association of other users and the user type prediction model constructed in advance predict the overdue of application user (i.e. the first user)
Probability, and whether be black list user based on overdue probability identification application user, it can be improved and determine whether application user is black name
The accuracy of single user, convenient for improving risk identification ability.
Specific embodiment
In order to make those skilled in the art more fully understand application scheme, below in conjunction in the embodiment of the present application
Attached drawing, the technical scheme in the embodiment of the application is clearly and completely described, it is clear that described embodiment is only
Some embodiments of the present application, instead of all the embodiments.Based on the embodiment in the application, those of ordinary skill in the art
Every other embodiment obtained without creative efforts, shall fall in the protection scope of this application.
The description and claims of this application and term " first " in above-mentioned attached drawing, " second " etc. are for distinguishing
Different objects, are not use to describe a particular order.In addition, term " includes " and " having " and their any deformations, it is intended that
It is to cover and non-exclusive includes.Such as the process, method, system, product or equipment for containing a series of steps or units do not have
It is defined in listed step or unit, but optionally further comprising the step of not listing or unit, or optionally also wrap
Include other step or units intrinsic for these process, methods, product or equipment.
Referenced herein " embodiment " is it is meant that a particular feature, structure, or characteristic described can wrap in conjunction with the embodiments
It is contained at least one embodiment of the application.Each position in the description occur the phrase might not each mean it is identical
Embodiment, nor the independent or alternative embodiment with other embodiments mutual exclusion.Those skilled in the art explicitly and
Implicitly understand, embodiment described herein can be combined with other embodiments.
Electronic equipment involved by the embodiment of the present application may include the various handheld devices with wireless communication function,
Mobile unit, wearable device calculate equipment or are connected to other processing equipments and various forms of radio modem
User equipment (user equipment, UE), mobile station (mobile station, MS), electronic equipment (terminal
Device) etc..For convenience of description, apparatus mentioned above is referred to as electronic equipment.The embodiment of the present application is carried out below detailed
It is thin to introduce.
Figure 1A is please referred to, the embodiment of the present application provides a kind of flow diagram of user type prediction technique.Specifically, such as
Shown in Figure 1A, a kind of user type prediction technique, comprising:
S101, electronic equipment construct user type prediction model when receiving predictions request.
In one implementation, which can be automatically generates depending on the user's operation, is also possible to
Period to be predicted, predetermined period can be one month, such as: when the end of month of every month reaches, prediction is sent to electronic equipment and is asked
It asks, that is to say, that when predetermined period reaches, the application user stored in presetting database is predicted, the application is implemented
Example is not construed as limiting this.
In one implementation, whether user type prediction model is black list user for predicting application user.On
State building user type prediction model include: based on presetting database determine training application user collect corresponding training dataset and
Training result collection, verifying application user collect corresponding validation data set and verification result collection;According to the training dataset and the instruction
Practice result set and constructs user type model of a syndrome to be tested;It is to be tested to the user type according to the validation data set and the verification result collection
Model of a syndrome is verified to obtain the user type prediction model.
Wherein, presetting database can be pre-stored in electronic equipment, alternatively, being stored in cloud server, electronics
Equipment obtains presetting database by access cloud server.Include in the presetting database from historical time to current time it
Between all application informations, such as: personal consumption installment contract, channel consumption installment contract, loan or loaning bill claim information
Deng;The presetting database may also include the related information obtained according to above-mentioned application information, such as: authorization message, refund information,
Instant communication information, other application informations etc., the embodiment of the present application does not limit this.
It should be noted that application user (such as: the first user) it can be applied to authorization object submission application business
Information, the application only focus on the user type of application user.
Wherein, application information include apply user identity information and association user (such as: the with the first user-association
Two users) identity information, association user can be spouse, relatives, friend etc.;Identity information includes user identifier, identity card
Number, telephone number, gender, age bracket, educational background, if blacklist etc., this application information further includes the amount of the loan, default when refunding
Long etc., the embodiment of the present application does not limit this.
The user identifier can be the pet name, identification card number or the telephone number for applying for user of application user, in this Shen
Please in embodiment, using the user identifier for applying for user as application user identifier (such as: the user identifier of the first user is first
User identifier), using the user identifier of association user as association user mark (such as: the user identifier of second user be second
User identifier), in the preset database to apply for the application information of user identifier storage application user and according to above-mentioned letter of application
Cease obtained related information.
Authorization message is used to indicate whether this application information authorizes, i.e., authorization message is authorization or unauthorized, in this Shen
The authorization of application user please can be determined according to the economic capability of the economic capability of authorized object or spouse, relatives in embodiment
Information, the embodiment of the present application do not limit this.
When authorization message is authorization, refund information is the refund temporal information of authorized object, or according to repayment period
Obtained normal refund or overdue refund etc. are limited, the embodiment of the present application does not limit this.
Instant communication information includes the dialogue in the message registration information for applying for user, short message record information and social application
Information or social circle's information etc., the embodiment of the present application does not limit this.
In one implementation, apply for that the identity information of user and association user can be application user in submission application
It inserts, can also be determined by the related data submitted in application information, in which: related data includes identity when business
Card, marriage certificate, property ownership certificate, residence booklet information etc., the embodiment of the present application does not limit this.
In one implementation, training application user collection includes that user is applied in multiple training, and training dataset is that this is more
The feature and its characteristic value of user's association user corresponding with its, training result collection are applied in each training in a training application user
For the corresponding refund information of training application user each in multiple training application user.
Verifying application user's collection includes that user is applied in multiple verifyings, and validation data set is that multiple verifying applies for every in user
The feature and its characteristic value of one verifying application user association user corresponding with its, verification result collection are that multiple verifying application is used
The corresponding refund information of user is applied in each verifying in family.
In one implementation, above-mentioned to determine that training application user collects corresponding training data based on the presetting database
Collection and training result collection, verifying application user collect corresponding validation data set and verification result collection include: the electronic equipment according to
Pre-set predetermined period, refund duration and the current time determine the first period, the second period, third period and when the 4th
Section;In the presetting database, first period is obtained respectively, the application user of the second period authorized application information obtains
One user collection and second user collection;In the presetting database, obtained at the beginning of from historical time to first period respectively
Between between, at the beginning of from the historical time to second period authorized application information application user and apply with it
The association user of user-association obtains third user collection and fourth user collection;Determine that first user collection collects it with third user
Between the user that is overlapped belong to the user being overlapped between training application user collection, the second user collection and the fourth user collection and belong to
Verifying application user collection;In the presetting database, determine at the beginning of from the historical time to first period respectively
Between the training application user collection, at the beginning of from the historical time to second period verifying application user concentrate
Each corresponding feature of application user and its corresponding characteristic value obtain the training dataset and the validation data set;It is default at this
In database, determine in the third period respectively with training application user collection, in the 4th period with verifying application user
The corresponding refund information of each application user is concentrated to obtain the training dataset and the verification result collection.
Wherein, predetermined period is it has been observed that details are not described herein;Historical time is first authorized application in presetting database
The information corresponding time;Current time is to receive the time of predictions request;The refund rule of a length of authorized application information when refund
It fixes time, such as: 3 months, 6 months etc.;First period is for determining training application user's collection;Second period is for determining verifying
Apply for user's collection;The third period includes the refund duration corresponding period after the first period, and the third period is for determining instruction
Practice result set;4th period included the refund duration corresponding period after the second period, and the 4th period is for determining verifying
Result set.
For example, as shown in Figure 1B, it is assumed that current time is on December 31st, 2017, and predetermined period is 1 month, is refunded
Shi Changwei 6 months, then determine that the first period was in May, 2017 according to the predetermined period, refund duration and current time, first uses
Family collection is the application user of the first period authorized application information;The third period includes that the refund duration after the first period is corresponding
Period, i.e. on November 30, in -2017 years on the 1st May in 2017;Third user integrates to be authorized as historical time on April 30th, 2017
Application information applies for user and the association user with application user-association.It include 10,000 application users when the first user collects, the
Three users collection includes 100,000,000 users, and applies for that the user being overlapped between user is 1,000 with 10,000 in 100,000,000 users, then
Training application user's collection is 1,000 application users, is determined based on the presetting database from historical time to April 30 in 2017
The feature and its corresponding characteristic value of each application user obtain the training dataset, base in this 1,000 application users between day
It determines in the presetting database in the third period, the refund information of each application user is somebody's turn to do in this 1,000 application users
Training result collection.
Determine that the second period was in June, 2017 according to the predetermined period, refund duration and current time, second user collection is
The application user of second period authorized application information;4th period differed a refund duration with the second period, then the 4th period
For the refund duration corresponding period after the second period, i.e. on December 31, in -2017 years on the 1st July in 2017;Fourth user collection
For the association user of the application user of historical time to authorized application information on May 31st, 2017 and its application user-association.When
Second user collection includes 10,000 apply users, and fourth user collection includes 100,000,000 1 ten thousand users, and with 1 in 100,000,000 1 ten thousand users
Ten thousand are applied for that the user being overlapped between users is 2,000, then verifying application user's collection is 2,000 application users, and being based on should
Presetting database is determined from historical time to each application user is corresponding in this 2,000 application users on May 31st, 2017
Feature and its corresponding characteristic value obtain the validation data set, based on the presetting database determine in four periods, this 2,000
The refund information of each application user obtains the verification result collection in a application user.
It is appreciated that determining the first period, second according to pre-set predetermined period, refund duration and the current time
Period, third period and the 4th period are based respectively on the first time period and the second time period in the preset database and determine training application
User's collection and verifying application user's collection, further concentrate each Shen according to training application user collection and verifying application user respectively
Please the corresponding feature of user and its corresponding characteristic value obtain training dataset and validation data set, respectively according in the third period
Application user collection refund information corresponding with the verifying application user each application user of concentration in the 4th period is trained to be trained
Result set and verification result collection.That is, obtaining training data or verify data since historical time, and selected instruction
Practice data and verify data and current time is close, the accuracy of user type prediction model can be improved.
In one implementation, above-mentioned to be verified according to the training dataset and training result collection building user type
Model includes: to obtain the training data each corresponding feature of training data and its corresponding characteristic value is concentrated to obtain multiple training
Feature set;Classified to obtain user type model of a syndrome to be tested according to multiple training characteristics collection and the training result collection.
Wherein, without limitation for sorting algorithm, it is special to multiple training that logistic regression or decision Tree algorithms can be used
Collection and the training result collection are classified, to obtain user type model of a syndrome to be tested.
In simple terms, user type model of a syndrome to be tested is equivalent to a function, each training characteristics is a constant, often
One training characteristics can obtain the corresponding training result of the training characteristics multiplied by a parameter.
It is appreciated that obtaining the corresponding feature of each training data and its corresponding characteristic value obtains multiple training characteristics
Collection, is classified to obtain user type model of a syndrome to be tested further according to multiple training characteristics collection and training result collection, use can be improved
The accuracy of family type model of a syndrome to be tested.
In one implementation, above-mentioned to be verified to the user type according to the validation data set and the verification result collection
Model is verified to obtain user type prediction model to include: to obtain the verify data to concentrate the corresponding feature of each verify data
And its corresponding characteristic value obtains multiple verifying feature sets;It is trained according to multiple verifying feature set and the verification result collection
Obtain user type model of a syndrome to be tested.
Wherein, without limitation for training algorithm used by verifying, common training method in neural network can be used,
Such as: gradient descent method (Gradient descent), Newton's algorithm (Newton ' s method), conjugate gradient method
(Conjugate gradient), quasi- Newton method (Quasi-Newton method), the least squares method Levenberg- to decay
Marquardt algorithm etc..
For example with gradient descent method, by verify data input user type model of a syndrome to be tested obtain one it is overdue general
Overdue probability verification result corresponding with the verify data is matched, if successful match, inputs next verifying by rate
Otherwise data obtain error function according to the matching and carry out reversed operation, so as to adjust user type model of a syndrome to be tested, and
The last one verify data successful match or reversed operation obtain user type prediction model after terminating, thus to user type
Model of a syndrome to be tested is verified, and the accuracy of user type prediction model is improved.
It is appreciated that determining that training application user collects corresponding instruction based on presetting database when receiving predictions request
Practice data set and training result collection, verifying application user collect corresponding validation data set and verification result collection, according to training data
Collection and training result collection construct user type model of a syndrome to be tested, then according to validation data set and verification result collection to user type
Model of a syndrome to be tested is verified to obtain user type prediction model.I.e. the application can obtain user by trained and verification method
Type prediction model, and whether be that black list user predicts to authorized application user according to the prediction model, it can be improved pre-
Survey application user whether be black list user accuracy, convenient for improve risk identification ability.
Further, above-mentioned building user type prediction model includes: to determine that training application is used based on the presetting database
The corresponding training dataset of family collection and training result collection, the first verifying application user collect corresponding first verification data collection and first
Verification result collection, the second verifying application user collect corresponding second validation data set and the second verification result collection;According to the training
Data set and the training result collection construct user type model of a syndrome to be tested;According to the first verification data collection and the first verifying knot
Fruit collection verifies user type model of a syndrome to be tested to obtain user type verifying model;According to second validation data set and
The second verification result collection is verified to obtain user type prediction model to user type verifying model.
Wherein, above-mentioned to determine that training application user collects corresponding training dataset and training result based on the presetting database
Collection, the first verifying application user collect corresponding first verification data collection and the first verification result collection, second verifying application user's collection
Corresponding second validation data set and the second verification result collection include: the electronic equipment according to pre-set predetermined period, also
Money duration and the current time determine the first period, the second period, third period, the 4th period, the 5th period and the 6th period;
In the presetting database, the application for obtaining first period, second period, the 5th period authorized application information respectively is used
Family obtains the first user collection, second user collection and the 5th user collection;In the presetting database, obtain respectively from historical time to
Between at the beginning of first period, at the beginning of from the historical time to second period, from the historical time
The application user of authorized application information and the association user with this application user-association between at the beginning of the 5th period
Obtain third user collection, fourth user collection and the 6th user collection;It determines and is overlapped between the 5th user collection and the 6th user collection
User belong to the user that is overlapped between training application user collection, first user collection and third user collection belong to this first
The user being overlapped between verifying application user collection, the second user collection and the fourth user collection belongs to second verifying application user
Collection;In the presetting database, the training application at the beginning of from the historical time to the 5th period is determined respectively
User's collection, at the beginning of from the historical time to first period the first verifying application user collection, from the history when
Between the second verifying application user between at the beginning of second period concentrate the corresponding feature of each application user and its
Corresponding characteristic value obtains the training dataset, the first verification data collection and second validation data set;In the preset data
Library determine in the 6th period respectively with training application user collection, in the third period with first verifying application user collection,
Refund information corresponding with the second verifying application user each application user of concentration obtains the training data in 4th period
Collection, the first verification result collection and the second verification result collection.
Wherein, the 5th period is for determining training application user's collection;First period is for determining the first verifying application user
Collection;Second period is for determining that the second verifying application user collects;6th period included the refund duration correspondence after the 5th period
Period, for determining training result collection;The third period includes the refund duration corresponding period after the first period, for true
Fixed first verification result collection;4th period included the refund duration corresponding period of the second period, for determining the second verifying knot
Fruit collection.
For example, as shown in Figure 1 C, it is assumed that current time is on December 31st, 2017, and predetermined period is 1 month, is refunded
Shi Changwei 6 months, then determine that the 5th period was in April, 2017 according to the predetermined period, refund duration and current time, the 5th uses
Family collection is the application user of the 5th period authorized application information;When 5th period included that the refund duration of the 6th period is corresponding
Section, on October 30, in as -2017 years on the 1st May in 2017;6th user integrates as historical time to authorization on March 31st, 2017 Shen
Please information application user and with application user-association association user.When the 5th user collection include 10,000 application users, the 6th
User's collection includes 9,000 users, and applies for that the user being overlapped between user is 900 with 10,000 in 9,000 users, then instructs
Practicing application user's collection is 1,000 application users, is determined based on the presetting database from historical time on March 31st, 2017
Between in this 1,000 application users the feature and its characteristic value of each application user obtain the training dataset, it is default based on this
Database determines that in the 6th period, the refund information of each application user obtains the training result in this 900 application users
Collection.
Determine that the first period was in May, 2017 according to the predetermined period, refund duration and current time, the first user collection is
The application user of first period authorized application information;The third period includes the refund duration corresponding period of the first period, as
On November 30, -2017 years on the 1st June in 2017;Third user integrates as historical time to authorized application information on April 30th, 2017
Apply for user and the association user with application user-association.When the first user collection includes 10,000 application users, third user Ji Bao
100,000,000 users are included, and apply for that the user being overlapped between user is 1,000 with 10,000 in 100,000,000 users, then the first verifying Shen
Please user collection be 1,000 application users, based on the presetting database determination from historical time on April 30th, 2017
The feature and its corresponding characteristic value of each application user obtains first verification data collection in this 1,000 application users, and being based on should
Presetting database determine in the third period, this 1,000 application users in it is each application user refund information obtain this first
Verification result collection.
Determine that the second period was in June, 2017 according to the predetermined period, refund duration and current time, second user collection is
The application user of second period authorized application information;4th period included the refund duration corresponding period of the second period, as
On December 31, -2017 years on the 1st July in 2017;Fourth user integrates as historical time to authorized application information on May 31st, 2017
Apply for user and the association user with application user-association.When second user collection includes 10,000 application users, fourth user Ji Bao
It includes 100,000,000 1 ten thousand users, and applies for that the user that is overlapped is 2,000 between users 10,000 in 100,000,000 1 ten thousand users, then second
Verifying application user's collection is 2,000 application users, is determined based on the presetting database from historical time to May 30 in 2017
Each corresponding feature of application user and its corresponding characteristic value obtain the second verifying number in this 2,000 application users between day
It according to collection, is determined in the 4th period based on the presetting database, the refund letter of each application user in 2,000 application users
Breath obtains the second verification result collection.
It is appreciated that the above method obtains user type prediction model by once training and twice verifying, and according to this
Whether prediction model is that black list user predicts to authorized application user, can further improve prediction application user whether be
The accuracy of black list user, convenient for improving risk identification ability.
S102, the electronic equipment are used based on presetting database determination with the first user and with the second of first user-association
Family.
In one implementation, which obtains from historical time to current time in the presetting database
Between authorized application information user obtain prediction application user collection, the first user be prediction application user concentration any one
Apply for user.
Wherein, above-mentioned authorized application information, the i.e. authorization message of this application information are authorization.That is, the first user
Until from historical time to current time, one of the user of authorized application information.
In one implementation, first user mark of the electronic equipment in the presetting database based on first user
Know the instant communication information and/or application information for searching first user.
Wherein, the first user identifier can be identification card number, be also possible to telephone number, can also be that pet name etc. has only
The identification information of one property.Electronic equipment can by and the information of the first user-association be stored in column corresponding with the first user identifier
In table, to improve the accuracy and efficiency for searching the first user related information.
It is appreciated that the determination of instant communication information and/or application information and first user-association based on the first user
Second user, improve the accuracy of determining association user.And the relevant information of association user also is stored in the preset data
In library, convenient for improve according to association user data prediction application user whether be black list user accuracy.
In one implementation, electronic equipment can include instant communication information and the Shen of application user identifier according to this
Please information, building comprising application user identifier knowledge mapping, the knowledge mapping have recorded with application user have incidence relation
Association user user identifier.
In another implementation, the electronic equipment is in the presetting database based on the first user of first user
The second user of identifier lookup and first user-association.That is, electronic equipment can will be closed with the first user as the aforementioned
The information of connection is stored in list corresponding with the first user identifier, which may include second with first user-association
User, to improve the accuracy and efficiency for searching second user.In another implementation, electronic equipment can basis
The determining second user with first user-association of knowledge mapping.Specifically, electronic equipment can obtain in the preset database
The knowledge mapping, the vertex of the knowledge mapping include user identifier or telephone number;And in the knowledge mapping, determine and first
At least one associated representative points of the corresponding vertex of user identifier, the representative points by meet preset quantity side and this
The corresponding vertex of one user identifier is connected;The corresponding user of user identifier that each representative points include is determined as and this application
The association user of user-association.
For the schematic diagram of the knowledge mapping shown in Fig. 1 D, wherein first, second, third, fourth, oneself be user identifier, A, B, C,
D, E is telephone number, and when applying for user identifier is first, the corresponding associated representative points in vertex of application user identifier can be
Second, fourth, oneself and third (i.e. association user can for second, fourth, oneself and third), wherein second and fourth pass through 2 sides and are connected with first, oneself
It is connected by 3 sides with first with third.
It should be noted that knowledge mapping shown in Fig. 1 D is only used for illustrating, the restriction to the embodiment of the present invention is not constituted.
In other feasible implementations, knowledge mapping can also include other users mark and telephone number, it is to be understood that
It can also be connected by the side of 4,5,8 or other quantity between two user identifiers in knowledge mapping, the embodiment of the present invention
This is not construed as limiting.
S103: the electronic equipment obtains the degree of association of first user and the second user based on the presetting database and is somebody's turn to do
The feature of second user and its corresponding the First Eigenvalue.
In one implementation, the degree of association may be greater than 0 number, and the degree of association can be used to indicate that application user
Intimate degree between (the first user) and association user (second user), the degree of association is bigger, indicate application user be associated with use
Intimate degree between family is higher.
In one implementation, electronic equipment can store being associated with for the first user and second user in the preset database
The incidence relation of relationship and other users and other associated association users of other users storage in the preset database same
In one incidence relation table, wherein other users are different from application user (and association user).The electronic equipment can will also be applied
The incidence relation of user and association user, other users and the incidence relation difference with other associated association users of other users
It is stored in different incidence relation tables, the embodiment of the present invention is not construed as limiting this.
In one implementation, there is the number of the second user of incidence relation can be 1 or more with the first user
It is a.When the quantity of second user is multiple, the degree of association between each second user and first user is mutually indepedent, i.e., respectively
The degree of association between a second user and first user may be the same or different.For example, the first user and second user
Incidence relation table can be as shown in table 1 below:
Table 1
First user identifier |
Second user mark |
The degree of association |
Zhang San |
Li Si |
0.2 |
Zhang San |
King five |
0.3 |
In another implementation, the degree of association can be determined by applying for the relationship of user's filling in application information,
As shown in table 2 below, if the relationship between the first user and second user is man and wife, the degree of association 3;If the first user and second
Relationship between user is relatives, then the degree of association is 2;If the relationship between the first user and second user is friend, it is associated with
Degree is 1.
Table 2
Relationship |
Man and wife |
Relatives |
Friend |
The degree of association |
3 |
2 |
1 |
In another implementation, the degree of association can be true by the immediate communication information between the first user and second user
It is fixed, such as: talk times are more, and the duration of call is longer, and the degree of association is bigger.
In another implementation, the electronic equipment be based on the instant communication information and/or application information obtain this
The degree of association of one user and the second user.
It is appreciated that the instant communication information and/or application information that are stored from presetting database obtain the first user and
The degree of association of second user, improves the accuracy of the determining degree of association, to improve the data prediction application according to association user
User whether be black list user accuracy.
In another implementation, if electronic equipment determines the pass of first user and second user according to knowledge mapping
Connection degree.If the corresponding vertex of the first user passes through the side vertex phase corresponding with association user of preset quantity in knowledge mapping
Even, then the preset quantity is smaller, and the degree of association of the first user and association user is bigger, and similarly, the preset quantity is bigger, and first uses
The degree of association of family and association user is smaller.
For the schematic diagram of the knowledge mapping shown in Fig. 1 D, when the first user identifier is first, the first user identifier pair
The associated representative points in the vertex answered can be second, fourth, oneself and third (i.e. association user can be second, fourth, oneself and third), wherein
Second and fourth pass through 2 sides and are connected with first, oneself and third be connected with first by 3 sides, at this point, the association of first and second (or fourth)
Degree can be 0.6, and the degree of association of first and oneself (or third) can be 0.4.
In one implementation, electronic equipment can be according to first degree of association and second of the first user and second user
The degree of association obtains the degree of association of the first user and second user.Specifically, electronic equipment can close first degree of association and second
The product of connection degree is determined as the degree of association of the first user and second user.Wherein, first degree of association can be according to aforementioned default
What quantity obtained.In one implementation, each side in knowledge mapping has weight, and electronic equipment can be according to knowledge graph
The weight on side associated with the first user, determines second degree of association in spectrum.
In one implementation, electronic equipment according to the determining association user with first user-association of knowledge mapping it
Afterwards, following steps can also be performed: determining at least one target line set corresponding with the first user identifier in knowledge mapping,
The target line set includes the side of aforementioned preset quantity, and the corresponding vertex of the first user identifier and preceding aim vertex pass through the mesh
All sides marked in line set are connected;For each target line set, the weight according to all sides in the target line set is true
Fixed second degree of association.
For the schematic diagram of the knowledge mapping shown in Fig. 1 D, wherein the number beside side in knowledge mapping is this
The weight on side, when the first user identifier be first when, the associated representative points in the corresponding vertex of the first user identifier be second, fourth, oneself
With third (i.e. association user can for second, fourth, oneself and third), wherein second and fourth pass through 2 sides and are connected with first, oneself leads to third
3 sides are crossed to be connected with first.At this point, first degree of association of first and second (perhaps fourth) can for 0.6 first and oneself (or third) first
The degree of association can be 0.4, and second degree of association of first and second can be 0.4*0.4=0.16, and second degree of association of first and fourth can be with
For 0.2*1=0.2, first and oneself second degree of association can be 0.3*0.7*1=0.21, and first and third second degree of association can be
0.2*0.3*1=0.06, therefore, the degree of association of first and second can be 0.6*0.16=0.096, and the degree of association of first and fourth can be
0.6*0.2=0.12, first and oneself degree of association can be 0.4*0.21=0.084, and first and third degree of association can be 0.4*
0.06=0.024.
In one implementation, the feature of second user includes plurality of classes, such as: gender, age bracket, educational background are
No blacklist etc., the corresponding characteristic value of each classification, and the corresponding the First Eigenvalue of inhomogeneity another characteristic is mutually indepedent.
For example, when the feature of second user includes applying for the gender of user, age bracket and this academic 3 classifications, and second
When the gender of user is male, age bracket is the middle age, educational background is master, electronic equipment obtains after handling the feature of second user
The corresponding the First Eigenvalue of the gender of the second user arrived can be 0.1, corresponding firstth spy of the age bracket of second user
Value indicative can be 0.2, and the corresponding the First Eigenvalue of educational background of second user can be 0.1.
In one implementation, the feature of second user and its corresponding the First Eigenvalue may include whether normally also
It is money, whether overdue and at least one of whether authorize, and its corresponding value.Wherein, unauthorized corresponding with authorization, it is unauthorized
Refer to and is not reviewed when user submits credit information by being rejected loans.For example, the feature of second user and its corresponding first
Characteristic value can be as shown in table 3 below:
Table 3
First user identifier |
Whether normally refund |
It is whether overdue |
Whether authorize |
Zhang San |
0 |
1 |
0 |
It should be noted that in the feature of second user, if in the normal value refunding, is whether overdue, whether authorizing only
Having a value is 1, that is to say, that if second user belongs to normal refund, any one situation in overdue and authorization, remaining
The value of two kinds of situations is 0.
In one implementation, the feature of second user can be one in normal refund, overdue and authorization.Example
Such as, if the overdue record of second user is as shown in table 3, the feature of second user is overdue refund, and the spy of second user
Levying corresponding the First Eigenvalue is 1.
Further, which obtains the spy of the second user based on the instant communication information and/or application information
Sign and its corresponding the First Eigenvalue.That is, storage instant communication information and/or application information obtain from presetting database
Take second user feature and its corresponding the First Eigenvalue, improve the accurate of determining feature and its corresponding the First Eigenvalue
Property, thus improve according to association user data prediction application user whether be black list user accuracy.
In one implementation, the above-mentioned feature and its corresponding that the second user is obtained based on the presetting database
One characteristic value include: based on the presetting database determine the corresponding feature in the corresponding community of the second user and the community and
Its corresponding characteristic value;Using the corresponding feature in the community and its corresponding characteristic value as the feature of the second user and its correspondence
The First Eigenvalue.
That is, obtaining multiple communities according to community's partitioning algorithm, then determine the corresponding overdue probability in each community, it will
The First Eigenvalue of the overdue probability of the community as it where second user.In the embodiment of the present application, community is divided
Algorithm without limitation, can be divided according to the geographical location of community, can also be according to the distance in the knowledge mapping of application user
It divides etc..
In another implementation, above-mentioned that the feature of the second user and its corresponding is obtained based on the presetting database
The First Eigenvalue includes: the second user identifier lookup second user pair in the presetting database, based on the second user
Answer multiple application informations;If the number of this application information be it is multiple, obtain the corresponding objective appraisal of each this application information be worth
To multiple objective appraisal values;The feature and its corresponding fisrt feature of the second user are obtained according to multiple objective appraisal value
Value.
That is, can obtain the corresponding target of each application information if the second user includes multiple application informations and comment
Value obtains multiple objective appraisal values, determines the feature of the second user and its corresponding first special according to multiple objective appraisal values
Value indicative.
In the embodiment of the present application, for obtaining the method for the corresponding objective appraisal value of each application information without limitation,
It can determine overdue dimension and the corresponding probability value of unauthorized dimension, it is then corresponding default according to overdue dimension and unauthorized dimension
Weight is weighted to obtain destination probability value.For determined according to multiple objective appraisal values the second user feature and its
Corresponding the First Eigenvalue also without limitation, can choose minimum value, maximum value or the weighted average in multiple objective appraisal values
Deng.
In one implementation, the feature of first user and its method of corresponding Second Eigenvalue are determined, can be joined
The method of feature and its corresponding the First Eigenvalue according to the determination second user described before, details are not described herein.
S104: the degree of association and the First Eigenvalue are input to the user type prediction model and obtained by the electronic equipment
The overdue probability of first user.
In one implementation, the value range of the overdue probability of the first user is [0,1], for describing application user
The probability of overdue refund.
It is appreciated that the degree of association and the First Eigenvalue of the first user and second user are input to user type prediction mould
Type obtains the overdue probability of the first user, i.e., the corresponding data characteristics of the first user is input to user type prediction model, can
Improve the accuracy for determining overdue probability.
In another implementation, this method further include: obtain the feature of first user and its corresponding second special
Value indicative;The degree of association and the First Eigenvalue are input to user type prediction model and obtain the overdue probability of first user by this
Include: by the degree of association, the First Eigenvalue and the Second Eigenvalue be input to the user type prediction model obtain this first
The overdue probability of user.
Wherein, method above-mentioned can refer to for the feature and its corresponding Second Eigenvalue that obtain the first user, herein
It repeats no more.It is appreciated that by the degree of association of the first user and second user, the Second Eigenvalue and second user of the first user
The First Eigenvalue be input to user type prediction model and obtain the overdue probability of the first user, i.e., by the corresponding number of the first user
It is input to user type prediction model according to feature, can further improve the accuracy for determining overdue probability.
In another implementation, this method further include: the quantity of the second user is multiple;This by the degree of association and
It includes: by each second use that the First Eigenvalue, which is input to user type prediction model and obtains the overdue probability of first user,
Family is input to the user type prediction model and obtains with the degree of association of first user, the First Eigenvalue of each second user
The overdue probability of first user.
It is appreciated that if receiving predictions request, based on presetting database determine the first user and with the first user-association
Multiple second users, then be based further on the degree of association that presetting database determines the first user Yu each second user, and
Obtain each second user feature and its corresponding the First Eigenvalue, then by the first user and each second user
The First Eigenvalue of the degree of association and second user input user type prediction model obtains the overdue probability of the first user.I.e.
The application can predict the first use according to the characteristic value of multiple other users (i.e. second user) and user type prediction model
The overdue probability at family, thus improve determine the first user whether be black list user accuracy, convenient for improve risk identification
Ability.
S105: the electronic equipment is based on the overdue probability and identifies that first user is black list user.
In the embodiment of the present application, whether first user, which is that the method for black list user is not done, is identified for the overdue probability
It limits, a settable targets threshold, when the overdue probability is greater than the targets threshold, determines first user for blacklist use
Family.
In method as shown in Figure 1A, when receiving predictions request, user type prediction model is constructed, is then based on
Presetting database determines the first user and the second user with the first user-association, then obtains the first user based on presetting database
With the feature and its corresponding the First Eigenvalue of the degree of association of second user and second user, then the first user is used with second
The degree of association at family and the First Eigenvalue input user type prediction model obtain the overdue probability of the first user, based on overdue
The first user of probabilistic forecasting is black list user.I.e. the application can according to the characteristic value of other users (i.e. second user), with
The degree of association of other users and the user type prediction model constructed in advance predict the overdue of application user (i.e. the first user)
Probability, and whether be black list user based on overdue probability identification application user, it can be improved and determine whether application user is black name
The accuracy of single user, convenient for improving risk identification ability.
Consistent with Figure 1A embodiment, referring to figure 2., the embodiment of the present application provides another user type prediction technique
Flow diagram.Specifically, as shown in Fig. 2, a kind of user type prediction technique, comprising:
S201, electronic equipment receive predictions request when, the electronic equipment be based on presetting database determine prediction application use
The corresponding predictive data set of family collection, training application user collect corresponding training dataset and training result collection, verifying application user
Collect corresponding validation data set and verification result collection.
S202, the electronic equipment construct user type model of a syndrome to be tested according to the training dataset and the training result collection.
S203, the electronic equipment according to the validation data set and the verification result collection to user type model of a syndrome to be tested into
Row verifying obtains the user type prediction model.
S204: the prediction data is concentrated each prediction data to be input to the user type prediction model and obtained by the electronic equipment
The corresponding overdue probability of each application user is concentrated to prediction application user.
S205: the electronic equipment is obtained black based on the corresponding overdue probability of each application user of prediction application user concentration
Name single user collection.
Wherein, the first user is any application user in presetting database.It is predicted for being determined based on presetting database
Apply for that user collects corresponding predictive data set, training applies for that user collects corresponding training dataset, verifying application user collects correspondence
Validation data set method, can refer to the second user that the first user and the first user-association are determined based on presetting database,
The feature of first user and the degree of association of the second user and the second user and its corresponding is determined based on presetting database again
The First Eigenvalue method, i.e., in data set corresponding data for its apply user-association association user and its corresponding spy
Value indicative.
In user type prediction technique as shown in Figure 2, when electronic equipment receives predictions request, it is based on present count
Determine that prediction application user collects corresponding predictive data set, training application user collects corresponding training data training result according to library
Collection, verifying application user collect corresponding validation data set and verification result collection, are tied according to above-mentioned training dataset and above-mentioned training
Fruit collection building user type model of a syndrome to be tested, then according to above-mentioned validation data set and above-mentioned verification result collection to above-mentioned user class
Type model of a syndrome to be tested is verified to obtain user type prediction model, concentrates each preset data to be input to above-mentioned prediction data
Above-mentioned user type prediction model obtains above-mentioned prediction application user and concentrates the corresponding overdue probability of each application user, then is based on
Above-mentioned prediction application user concentrates the corresponding overdue probability of each application user to obtain black list user's collection.I.e. the application can lead to
It crosses trained and verification method and obtains user type prediction model, and the black name in prediction application user is determined according to the prediction model
Single user, can be improved prediction application user whether be black list user accuracy, convenient for improve risk identification ability.
Consistent with the embodiment of Figure 1A and Fig. 2, referring to figure 3., Fig. 3 is a kind of user type provided by the embodiments of the present application
The structural schematic diagram of prediction meanss, the device are applied to electronic equipment.As shown in figure 3, above-mentioned user type prediction meanss 300 are wrapped
It includes:
Receiving unit 301 is for receiving predictions request;
Processing unit 302 is for constructing user type prediction model;Based on presetting database determine the first user and with institute
State the second user of the first user-association;The pass of first user and the second user are obtained based on the presetting database
The feature and its corresponding the First Eigenvalue of connection degree and the second user;The degree of association and the First Eigenvalue are inputted
The overdue probability of first user is obtained to the user type prediction model;Based on first described in the overdue probabilistic forecasting
User is black list user.
It is appreciated that processing unit 302 constructs user type prediction model if receiving unit 301 receives predictions request,
It is then based on presetting database and determines the first user and the second user with the first user-association, then obtained based on presetting database
The feature and its corresponding the First Eigenvalue of first user and the degree of association of second user and second user, then by the first user
The overdue probability of the first user is obtained with the degree of association of second user and the First Eigenvalue input user type prediction model,
It is black list user based on overdue the first user of probabilistic forecasting.I.e. the application can be according to the spy of other users (i.e. second user)
Value indicative predicts application user (i.e. the first user) with the degrees of association of other users and the user type prediction model constructed in advance
Overdue probability, and based on overdue probability identification application user whether be black list user, can be improved determine application user whether
For the accuracy of black list user, it is convenient for improving risk identification ability.In a possible example, in the building user class
In terms of type prediction model, the processing unit 302 is specifically used for determining training application user's collection pair based on the presetting database
Training dataset and training result collection, the corresponding validation data set of verifying application user's collection and the verification result collection answered;According to institute
State training dataset and training result collection building user type model of a syndrome to be tested;According to the validation data set and described test
Card result set verifies user type model of a syndrome to be tested to obtain the user type prediction model.
In a possible example, determine that training application user collects corresponding instruction based on the presetting database described
Practice data set and training result collection, verifying application user collect corresponding validation data set and verification result collection aspect, the processing
Unit 302 is specifically used for determining the first period, second according to pre-set predetermined period, refund duration and the current time
Period, third period and the 4th period;In the presetting database, first period, second period are obtained respectively
The application user of authorized application information obtains the first user collection and second user collection;In the presetting database, obtain respectively
At the beginning of from historical time to first period, at the beginning of from the historical time to second period
Between authorized application information application user and with its apply the association user of user-association obtain third user collection and the 4th use
Family collection;Determine the user being overlapped between first user collection and third user collection belong to the training application user collection,
The user being overlapped between the second user collection and the fourth user collection belongs to verifying application user's collection;Described default
In database, the training application user at the beginning of from the historical time to first period is determined respectively
Collection concentrates each application user from the historical time to verifying described the end time of first period application user
Or the associated corresponding feature of association user and its corresponding characteristic value obtain the training dataset and the verify data
Collection;In the presetting database, is determined in the third period respectively and apply for user's collection, the 4th period with the training
In with verifying application user concentrate the corresponding refund information of each application user, obtain the training dataset and described test
Demonstrate,prove result set.
In a possible example, it is described based on presetting database determine the first user and with first user close
In terms of the second user of connection, the processing unit 302 is specifically used in the presetting database, obtains from historical time to working as
The application user of authorized application information obtains prediction application user's collection between the preceding time, and first user is prediction application
Any one application user that user concentrates;In the presetting database, the first user identifier based on first user
Search the instant communication information and/or application information of first user;Based on the instant communication information and/or letter of application
Breath, the determining second user with first user-association.
In a possible example, it is described based on the presetting database obtain the second user feature and its
In terms of corresponding the First Eigenvalue, the processing unit 302 is specifically used in the presetting database, uses based on described second
Second user described in the second user identifier lookup at family corresponds to multiple application informations;If the number of the application information be it is multiple,
It obtains the corresponding objective appraisal value of each application information and obtains multiple objective appraisal values;According to the multiple objective appraisal value
Obtain the second user feature and its corresponding the First Eigenvalue.
In a possible example, it is described based on the presetting database obtain the second user feature and its
In terms of corresponding the First Eigenvalue, the processing unit 302 is specifically used for determining that described second uses based on the presetting database
The corresponding feature in the corresponding community in family and the community and its corresponding characteristic value;By the corresponding feature in the community and its right
Feature and its corresponding the First Eigenvalue of the characteristic value answered as the second user.
In a possible example, the quantity of the second user is multiple;Described by the degree of association and described
The First Eigenvalue is input to the user type prediction model and obtains the overdue probability aspect of first user, and the processing is single
Member 302 is specifically used for first by each second user and the degree of association of first user, each second user
Characteristic value is input to the user type prediction model and obtains the overdue probability of first user.
Consistent with the embodiment of Figure 1A and Fig. 2, referring to figure 4., Fig. 4 is a kind of electronic equipment provided by the embodiments of the present application
Structural schematic diagram.As shown in figure 4, the electronic equipment 400 includes processor 410, memory 420 and one or more programs
430, wherein said one or multiple programs 430 are stored in above-mentioned memory 420, and are configured by above-mentioned processor
410 execute, and above procedure 430 includes the instruction for executing following steps:
When receiving predictions request, user type prediction model is constructed;
The first user and second user with first user-association are determined based on presetting database;
Determine that the degree of association of first user and the second user and described second is used based on the presetting database
The feature at family and its corresponding the First Eigenvalue;
The degree of association and the First Eigenvalue are input to the user type prediction model and obtain first use
The overdue probability at family;
It is black list user based on the first user described in the overdue probabilistic forecasting.
It is appreciated that constructing user type prediction model if electronic equipment 400 receives predictions request, being then based on pre-
If database determines the first user and the second user with the first user-association, then based on presetting database obtain the first user with
The degree of association of second user and the feature of second user and its corresponding the First Eigenvalue, then by the first user and second user
The degree of association and the First Eigenvalue input user type prediction model obtain the overdue probability of the first user, based on overdue general
Rate predicts that the first user is black list user.I.e. the application can be according to the characteristic value of other users (i.e. second user) and its
The degree of association of his user and the user type prediction model constructed in advance predict the overdue general of application user (i.e. the first user)
Rate, and whether be black list user based on overdue probability identification application user, it can be improved and determine whether application user is blacklist
The accuracy of user, convenient for improving risk identification ability.
Instruction tool in a possible example, in terms of above-mentioned building user type prediction model, in the program 430
Body is for performing the following operations:
Determine that training application user collects corresponding training dataset and training result collection, verifying based on the presetting database
Apply for that user collects corresponding validation data set and verification result collection;
According to the training dataset and the training result collection, user type model of a syndrome to be tested is constructed;
According to the validation data set and the verification result collection, user type model of a syndrome to be tested verify
To the user type prediction model.
In a possible example, determine that training application user collects corresponding instruction based on the presetting database above-mentioned
Practice data set and training result collection, verifying application user collect corresponding validation data set and verification result collection aspect, the program 430
In instruction be specifically used for executing following operation:
The first period, the second period, are determined according to pre-set predetermined period, refund duration and the current time
Three periods and the 4th period;
In the presetting database, the Shen of first period, the second period authorized application information are obtained respectively
Please user obtain the first user collection and second user collection;
In the presetting database, obtain at the beginning of from historical time to first period respectively, from
The application user of authorized application information and apply for user with it between at the beginning of the historical time to second period
Associated association user obtains third user collection and fourth user collection;
Determine that the user being overlapped between the first user collection and third user collection belongs to the training application user
The user being overlapped between collection, the second user collection and the fourth user collection belongs to verifying application user's collection;
In the presetting database, determined at the beginning of from the historical time to first period respectively
The training application user collection applies for user from the historical time to verifying described the end time of first period
Each application user or the corresponding feature of associated association user and its corresponding characteristic value is concentrated to obtain the training data
Collection and the validation data set;
In the presetting database, determine respectively in the third period with the training application user collection, described the
Concentrate in four periods with verifying application user the corresponding refund information of each application user obtain the training dataset and
The verification result collection.
In a possible example, it is above-mentioned based on presetting database determine the first user and with first user close
In terms of the second user of connection, the instruction in the program 430 is specifically used for executing following operation:
In the presetting database, the application user of the authorized application information between historical time to current time is obtained
Prediction application user's collection is obtained, first user is any one application user that the prediction application user concentrates;
In the presetting database, the first user identifier based on first user search first user i.e.
When communication information and/or application information;
Based on the instant communication information and/or the determining second user with first user-association of application information.
In a possible example, it is above-mentioned based on the presetting database obtain the second user feature and its
In terms of corresponding the First Eigenvalue, the instruction in the program 430 is specifically used for executing following operation:
In the presetting database, second user described in the second user identifier lookup based on the second user is corresponding
Multiple application informations;
If the number of the application information be it is multiple, obtain the corresponding objective appraisal value of each application information obtain it is more
A objective appraisal value;
According to the multiple objective appraisal value obtain the second user feature and its corresponding the First Eigenvalue.
In a possible example, it is above-mentioned based on the presetting database obtain the second user feature and its
In terms of corresponding the First Eigenvalue, the instruction in the program 430 is specifically used for executing following operation:
Based on the presetting database, the corresponding feature in the corresponding community of the second user and the community is determined
And its corresponding characteristic value;
Using the corresponding feature in the community and its corresponding characteristic value as the feature of the second user and its corresponding
The First Eigenvalue.
In a possible example, the quantity of the second user is multiple;Above-mentioned by the degree of association and described
The First Eigenvalue is input to the user type prediction model and obtains the overdue probability aspect of first user, the program 430
In instruction be specifically used for executing following operation:
By the First Eigenvalue of each second user and the degree of association of first user, each second user
It is input to the user type prediction model and obtains the overdue probability of first user.
The embodiment of the present application also provides a kind of computer storage medium, wherein the computer storage medium is stored for depositing
Computer program is stored up, which makes computer execute either record part of method or complete in such as embodiment of the method
Portion's step, computer include electronic equipment.
The embodiment of the present application also provides a kind of computer program product, and computer program product includes storing computer journey
The non-transient computer readable storage medium of sequence, computer program are operable to execute computer as remembered in embodiment of the method
Some or all of either load method step.The computer program product can be a software installation packet, and computer includes
Electronic equipment.
It should be noted that for the various method embodiments described above, for simple description, therefore, it is stated as a series of
Combination of actions, but those skilled in the art should understand that, the application is not limited by the described action sequence because
According to the application, some steps may be performed in other sequences or simultaneously.Secondly, those skilled in the art should also know
It knows, the embodiments described in the specification are all preferred embodiments, related movement and mode not necessarily the application
It is necessary.
In the above-described embodiments, it all emphasizes particularly on different fields to the description of each embodiment, there is no the portion being described in detail in some embodiment
Point, reference can be made to the related descriptions of other embodiments.
In several embodiments provided herein, it should be understood that disclosed device, it can be by another way
It realizes.For example, the apparatus embodiments described above are merely exemplary, such as the division of unit, only a kind of logic
Function division, there may be another division manner in actual implementation, such as multiple units or components can combine or can collect
At another system is arrived, or some features can be ignored or not executed.Another point, shown or discussed mutual coupling
It closes or direct-coupling or communication connection can be through some interfaces, the indirect coupling or communication connection of device or unit can be with
It is electrical or other forms.
Unit may or may not be physically separated as illustrated by the separation member, shown as a unit
Component may or may not be physical unit, it can and it is in one place, or may be distributed over multiple networks
On unit.It can some or all of the units may be selected to achieve the purpose of the solution of this embodiment according to the actual needs.
It, can also be in addition, each functional unit in each embodiment of the application can integrate in one processing unit
It is that each unit physically exists alone, can also be integrated in one unit with two or more units.Above-mentioned integrated list
Member both can take the form of hardware realization, can also be realized in the form of software program mode.
If integrated unit is realized and when sold or used as an independent product in the form of software program mode, can
To be stored in a computer-readable access to memory.Based on this understanding, the technical solution of the application is substantially in other words
The all or part of the part that contributes to existing technology or the technical solution can embody in the form of software products
Come, which is stored in a memory, including some instructions are used so that a computer equipment (can be
Personal computer, server or network equipment etc.) execute each embodiment method of the application all or part of the steps.And it is preceding
The memory stated includes: USB flash disk, read-only memory (read-only memory, ROM), random access memory (random
Access memory, RAM), mobile hard disk, the various media that can store program code such as magnetic or disk.
Those of ordinary skill in the art will appreciate that all or part of the steps in the various methods of above-described embodiment is can
It is completed with instructing relevant hardware by program, which can store in a computer-readable memory, memory
It may include: flash disk, ROM, RAM, disk or CD etc..
The embodiment of the present application is described in detail above, specific case used herein to the principle of the application and
Embodiment is expounded, the description of the example is only used to help understand the method for the present application and its core ideas;
At the same time, for those skilled in the art can in specific embodiments and applications according to the thought of the application
There is change place, to sum up, the contents of this specification should not be construed as limiting the present application.