CN110517132A - Credit-graded approach, system, terminal and computer readable storage medium - Google Patents
Credit-graded approach, system, terminal and computer readable storage medium Download PDFInfo
- Publication number
- CN110517132A CN110517132A CN201910623505.6A CN201910623505A CN110517132A CN 110517132 A CN110517132 A CN 110517132A CN 201910623505 A CN201910623505 A CN 201910623505A CN 110517132 A CN110517132 A CN 110517132A
- Authority
- CN
- China
- Prior art keywords
- users
- grading
- relationship
- target user
- data
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
- G06Q40/03—Credit; Loans; Processing thereof
Landscapes
- Business, Economics & Management (AREA)
- Accounting & Taxation (AREA)
- Finance (AREA)
- Engineering & Computer Science (AREA)
- Development Economics (AREA)
- Economics (AREA)
- Marketing (AREA)
- Strategic Management (AREA)
- Technology Law (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a kind of credit-graded approach, system, terminal and computer readable storage medium, method includes: the initial credit scoring for obtaining target user and the abnormal reference scoring of associated other users；Associated other users are classified, the relationship grading of each other users is obtained, and obtain the corresponding preset weights of each relationship grading；According to the scoring of the abnormal reference of each other users, relationship grading and the corresponding preset weights of each relationship grading, obtain the reference overall score of the associated every other user of target user, and initial credit scoring is subtracted to the reference overall score of every other user, obtain the synthesis credit scoring of target user.The present invention by reference to the associated other users exception reference of target user and corresponding other users of addition classification, to be the credit scoring model of principle so as to model trustship, help solves at present that personal collage-credit data acquisition is not comprehensive enough, has the problem of limitation.
Description
Technical field
The present invention relates to computer field more particularly to credit-graded approach, system, terminal and computer-readable storage mediums
Matter.
Background technique
With the quick emergence of internet financial industry, the business such as net is borrowed, manages money matters on financial lease and line are received very
The concern and participation of more clients, but then, can user refund in time has been related to enterprise and borrower's financial asset
Safety, then many financial companies establish own reference score-system.And current reference points-scoring system is used in evaluation
When family reference situation is to determine user's accrediting amount, the borrower's personal information being collected into, but practical borrower are mainly relied on
The data area that personal information can refer to is relatively narrow, therefore credit investigation system scoring is more unilateral, and there are technical limitations.
Summary of the invention
The main purpose of the present invention is to provide a kind of credit-graded approach, system, terminal and computer-readable storage mediums
Matter, it is intended to which it is more unilateral to solve the scoring of current credit investigation system, there is technical issues that.
To achieve the above object, the present invention provides a kind of credit-graded approach, comprising the following steps:
Obtain the initial credit scoring of target user and the abnormal reference scoring of the associated other users of the target user；
The associated other users of the target user are classified, obtain the relationship grading of each other users, and obtain
The corresponding preset weights of each relationship grading are taken, wherein relationship grading is higher, and the corresponding preset weights obtained are bigger；
According to the scoring of the abnormal reference of each other users, relationship grading and the corresponding default power of each relationship grading
Value obtains the reference overall score of the associated every other user of the target user, and initial credit scoring is subtracted institute
The reference overall score for stating every other user obtains the synthesis credit scoring of the target user.
Optionally, described according to the scoring of the abnormal reference of each other users, relationship grading and the grading pair of each relationship
The preset weights answered, the step of obtaining the reference overall score of the associated every other user of the target user include:
According to the scoring of the abnormal reference of each other users, relationship grading and the corresponding default power of each relationship grading
Value establishes the mapping relations of other users, abnormal reference scoring and preset weights；
The mapping relations data correspondence of foundation is input in the arithmetic unit including formula P=∑ Wi × Qi, to export
The reference overall score of the associated every other user of target user is stated, wherein P is the associated every other user's of target user
Reference overall score, Wi are that the abnormal reference of associated i-th of the other users of the target user scores, and Qi is i-th of other use
The corresponding preset weights of relationship grading at family, i ∈ [1, n], n are the associated other users sum of the target user.
Optionally, it is described obtain target user initial credit scoring and the associated other users of the target user it is different
Chang Zhengxin score the step of include:
All kinds of economic activity data of target user are obtained, and all kinds of economic activity data are input to goal-selling
In user credit Rating Model, to obtain the initial credit scoring of the target user；
Obtain the corresponding economic activity data of the associated each other users of the target user, and from each other users
The corresponding abnormal economic activity data of the other users are extracted in corresponding economic activity data, other each use are obtained with statistics
The abnormal data amount at family；
According to the abnormal data amount of each other users, the default mapping between abnormal data amount and abnormal reference scoring is closed
It is that the corresponding abnormal reference scoring of each other users is inquired in table, wherein abnormal data amount is more, and corresponding exception reference is commented
Divide higher.
Optionally, described that the associated other users of the target user are classified, obtain the pass of each other users
Being the step of grading includes:
The historical sample interaction data between the target user and other users is obtained, and is gone through by clustering algorithm to described
History sample interaction data carries out clustering, to obtain multiple interaction gradings section；
The interaction data in preset time between the target user and each other users is obtained, according to the interactive number
According to the corresponding affiliated interaction grading section of each other users of determination；
By the corresponding preset relation grading in interaction grading section as each other users in affiliated interaction grading section
Relationship grading, wherein the corresponding endpoint value in interaction grading section is bigger, preset relation grading is higher.
Optionally, described that clustering is carried out to the historical sample interaction data by clustering algorithm, it is multiple to obtain
Interacting the step of grading section includes:
The object set of clustering algorithm is constructed by input gene expression matrix, the object set is interacted by all historical samples
The corresponding group of data points of data at；
K data point is chosen from the object set, wherein K is greater than or equal to 2, and with the K in the object set
A data point is that cluster centre establishes aggregate of data respectively；
Cluster iterative operation is carried out to by the aggregate of data of cluster centre of K data point, it is full with the aggregate of data after iteration
When sufficient preset termination condition, the corresponding aggregate of data of K cluster centre of newest an iteration is exported, and by the K of output
Aggregate of data range is as multiple interaction gradings section.
Optionally, described that the associated other users of the target user are classified, obtain the pass of each other users
After the step of system's grading, further includes:
The address book data of target terminal user is obtained, and is determined in the address book data with the presence or absence of other any use
The corresponding communication number in family；
When communication number corresponding there are any other users in the address book data, it is whole to obtain the target user
In the first history IP address and address book data at end there are the second history IP of the correspondence other users terminal of communication number
Location, and count identical IP address number in the first history IP address and the second history IP address；
When identical IP address number is greater than preset value, by the relationship of the corresponding other users of the second history IP address
Grading promotes a grading, and executes step: obtaining the corresponding preset weights of each relationship grading.
Optionally, it is described when identical IP address number be greater than preset value when, by the second history IP address it is corresponding its
His user relationship grading promoted a grading the step of include:
When identical IP address number is greater than preset value, the pass of the corresponding other users of the second history IP address is judged
Whether system's grading is highest grading；
If it is not, the relationship grading of the corresponding other users of the second history IP address is then promoted a grading.
In addition, to achieve the above object, the present invention also provides a kind of credit scoring system, the credit scoring system packet
It includes:
Module is obtained, the initial credit for obtaining target user scores and the target user associated other users
Abnormal reference scoring；
Diversity module obtains each other users for being classified to the associated other users of the target user
Relationship grading, and the corresponding preset weights of each relationship grading are obtained, wherein relationship grading is higher, the corresponding preset weights obtained
It is bigger；
The acquisition module is also used to be scored according to the abnormal reference of each other users, relationship is graded and each pass
The corresponding preset weights of system's grading obtain the reference overall score of the associated every other user of the target user, and will be described
Initial credit, which scores, subtracts the reference overall score of the every other user, obtains the synthesis credit scoring of the target user.
In addition, to achieve the above object, the present invention also provides a kind of terminal, the terminal includes: communication module, storage
Device, processor and it is stored in the computer program that can be run on the memory and on the processor, the computer journey
The step of sequence realizes credit-graded approach as described above when being executed by the processor.
In addition, to achieve the above object, it is described computer-readable the present invention also provides a kind of computer readable storage medium
Computer program is stored on storage medium, the computer program realizes credit scoring as described above when being executed by processor
The step of method.
A kind of credit-graded approach, system, terminal and the computer readable storage medium that the embodiment of the present invention proposes, pass through
Obtain the initial credit scoring of target user and the abnormal reference scoring of the associated other users of the target user；To the mesh
The other users of mark user-association are classified, and the relationship grading of each other users is obtained, and obtain each relationship grading pair
The preset weights answered, wherein relationship grading is higher, and the corresponding preset weights obtained are bigger；According to the exception sign of each other users
Letter scoring, relationship grading and the corresponding preset weights of each relationship grading, it is associated every other to obtain the target user
The reference overall score of user, and the initial credit is scored and subtracts the reference overall score of the every other user, obtain institute
State the synthesis credit scoring of target user.Wherein, with reference to and joined the reference situations of the associated other users of target user, and
Corresponding preset weights are graded by relationship to embody the how far of each other users Yu target user's social relationships, so that right
When target user carries out credit appraisal, it can add by the credit scoring model that joined associated other users reference situation
Personal reference acquires the comprehensive of data in by the credit scoring system that model trustship is principle by force, solves collage-credit data
There is technical limitation in acquisition.
Detailed description of the invention
Fig. 1 is the terminal structure schematic diagram for the hardware running environment that the embodiment of the present invention is related to；
Fig. 2 is the flow diagram of credit-graded approach first embodiment of the present invention；
Fig. 3 is the flow diagram of step S30 in credit-graded approach second embodiment of the present invention；
Fig. 4 is the flow diagram of step S10 in credit-graded approach 3rd embodiment of the present invention；
Fig. 5 is the flow diagram of step S20 in credit-graded approach fourth embodiment of the present invention；
Fig. 6 is the functional block diagram of one embodiment of credit scoring system of the present invention.
The embodiments will be further described with reference to the accompanying drawings for the realization, the function and the advantages of the object of the present invention.
Specific embodiment
It should be appreciated that the specific embodiments described herein are merely illustrative of the present invention, it is not intended to limit the present invention.
Fig. 1 is please referred to, Fig. 1 is the hardware structural diagram of terminal provided by the present invention.The terminal can be service
Device, can be the device end with credit scoring model, such as computer, and the terminal may include communication module 10, deposit
The components such as reservoir 20 and processor 30.In the terminal, the processor 30 respectively with the memory 20 and described
Communication module 10 connects, and is stored with computer program on the memory 20, the computer program is held by processor 30 simultaneously
Row, when the computer program executes the step of realization following methods embodiment.
Communication module 10 can be connect by network with external communications equipment.Communication module 10 can receive external communication and set
The request that preparation goes out, can also send request, instruction and information to the external communications equipment.The external communications equipment can be with
It is user terminal, other servers etc..
Memory 20 can be used for storing software program and various data.Memory 20 can mainly include storing program area
The storage data area and, wherein storing program area can application program needed for storage program area, at least one function (for example obtain
The initial credit of target user is taken to score) etc.；Storage data area may include database, and storage data area can be stored according to terminal
Use created data or information etc..In addition, memory 20 may include high-speed random access memory, it can also include non-
Volatile memory, for example, at least a disk memory, flush memory device or other volatile solid-state parts.
Processor 30 is the control centre of terminal, using the various pieces of various interfaces and the entire terminal of connection, is led to
It crosses operation or executes the software program and/or module being stored in memory 20, and call the number being stored in memory 20
According to, execute terminal various functions and processing data, thus to terminal carry out integral monitoring.Processor 30 may include one or more
A processing unit；Optionally, processor 30 can integrate application processor and modem processor, wherein application processor master
Processing operation system, user interface and application program etc. are wanted, modem processor mainly handles wireless communication.It is understood that
It is that above-mentioned modem processor can not also be integrated into processor 30.
Although Fig. 1 is not shown, above-mentioned terminal can also guarantee it for connecting to power supply including circuit control module
The normal work of his component.Above-mentioned terminal can also include display module, for extracting the data in memory 20, and show
The system interface of terminal scores with the credit scoring of the interactive interface of user and all types of user and/or reference.Art technology
Personnel are appreciated that the restriction of the not structure paired terminal of terminal structure shown in Fig. 1, may include more more or less than illustrating
Component, perhaps combine certain components or different component layouts.
Based on above-mentioned hardware configuration, each embodiment of the method for the present invention is proposed.
Referring to fig. 2, in the first embodiment of credit-graded approach of the present invention, which comprises
Step S10 obtains the initial credit scoring and the exception of the associated other users of the target user of target user
Reference scoring；
Wherein, target user is the user for needing to carry out credit or reference scoring, can obtain the various of target user
The data information of dimension, and existing reference Rating Model is inputted, to score the score value of output as initial credit.Or
Basic score value can be set for target user, then basic score value is added by the data information of code of points and target user
Subtract, initial credit scoring is obtained with this.It should be noted that the data information of each dimension of target user, what is only related to is target
The associated data that user voluntarily generates, such as the IP address of target user, fund account remaining sum buy social insurance situation
Etc..
The quantity of the associated other users of target user can be fixed, and the determination of associated other users can be use
Other users of the family after through terminal transacting business and fill data, in the system data that user fills in based on the received
What information determined, the other users information may include ID card No., at least one of cell-phone number and name.May be used also
To be determined by the social networks of target user, such as it is divided into lineal family members, relatives, friend and colleague etc. from closely to remote
Social relationships rank.Further, the abnormal reference scoring of other users can be scored with the initial credit of reference target user
Consistent code of points is determined, and can also only consider that part abnormal in each reference dimension of target user is determined, example
Such as, it can be determined by the data bulk of associated other users exception or the dimension being related to.
Step S20 is classified the associated other users of the target user, and the relationship for obtaining each other users is commented
Grade, and the corresponding preset weights of each relationship grading are obtained, wherein relationship grading is higher, and the corresponding preset weights obtained are bigger；
The distance for being referred to social relationships between target user and other users is classified, when the quantity of target user
It is fixed and is that can choose the other users of the nearest fixed quantity of social relationships when determining by social networks as pass
The other users of connection, can also first select every other user screen again the fixed quantity finally needed it is associated other
User.Wherein can use target user and the frequency that contacts of each other users be determined, can be combined with household register data,
The information such as home address are determined.Wherein relationship grading is higher indicates closer with the social relationships of target user.It needs to illustrate
, social relationships are closer, and the degree of association of practical other users and target user are higher, participate in financial business and generate money
A possibility that producing transaction is higher, therefore the corresponding weight that can be graded by setting relationship is embodied, each relationship grading
Corresponding weight is not identical, and the weight of the corresponding required setting in for target user's credit scoring is low compared to relationship grading
Other users want higher.
It should be noted that processor in advance carry out each relationship grade corresponding preset weights setting when, can be with
Setting fixed multiplying power/consistent difference, two neighboring relationship is graded corresponding preset weights, a consistent difference can be differed,
It can be between the corresponding preset weights of two neighboring relationship grading, the preset weights and relativeness that relativeness is graded high are commented
Multiple between the low preset weights of grade is fixed multiplying power, and wherein the fixed multiplying power is greater than 1.Optionally, function can also be passed through
Relationship carries out each relationship and grades the autonomous settings of corresponding preset weights, such as all relationships corresponding preset weights of grading are in
Index ascendant trend.As long as it is higher totally to meet relationship grading, the corresponding preset weights obtained are bigger, physical relationship grading
The rule of corresponding preset weights setting a variety of can consider.
Step S30, it is corresponding according to the scoring of the abnormal reference of each other users, relationship grading and the grading of each relationship
Preset weights obtain the reference overall score of the associated every other user of the target user, and the initial credit are scored
The reference overall score for subtracting the every other user obtains the synthesis credit scoring of the target user.
In the present embodiment, other users and the correlation degree of target user are characterized by the weight of grading, and then will
Weight combines the abnormal reference of corresponding other users to score the reference overall scores of available associated other users.Wherein calculate
The method of reference overall score, which can be, establishes mapping table by above- mentioned information, and all of fixed quantity will be found from mapping table
The abnormal reference scoring of each other users is multiplied simultaneously with the corresponding set preset weights of belonging relation grading in other users
Reference overall score of the accumulated result as other users.Alternatively, screen out for the lower other users of relationship grading
To the associated other users of fixed quantity, then the mean value that the abnormal reference of every other user in belonging relation grading is scored
Reference overall score multiplied by the sum being added after relationship grading and corresponding weight, as other users.
It is mesh respectively it is understood that the credit scoring of target user is combined two aspects in the present embodiment
The reference situation of user itself is marked, and the reference situation for finding associated other users is borrowed as reference to look for another way
The reference situation for having helped the other users of association social relationships, the credit for expanding target user collect data volume and dimension, solution
The relatively narrow problem of target user's (i.e. practical business applicant, such as borrower) credit data range of having determined, so that credit investigation system
Rating Model breaches technical limitation, and credit scoring result is very comprehensive.
The present embodiment passes through the initial credit scoring and the associated other users of the target user for obtaining target user
Abnormal reference scoring；The associated other users of the target user are classified, the relationship grading of each other users is obtained,
And the corresponding preset weights of each relationship grading are obtained, wherein relationship grading is higher, and the corresponding preset weights obtained are bigger；According to
The scoring of abnormal reference, relationship grading and the corresponding preset weights of each relationship grading of each other users, obtain the mesh
The reference overall score of the every other user of user-association is marked, and initial credit scoring is subtracted into the every other user
Reference overall score, obtain the synthesis credit scoring of the target user.Wherein, with reference to and joined target user it is associated its
The reference situation of his user, and corresponding preset weights are graded to embody each other users and target user society pass by relationship
The how far of system can be by joined associated other users reference feelings when so that carrying out credit appraisal to target user
The credit scoring model of condition reinforces the personal reference in by the credit scoring system that model trustship is principle and acquires the complete of data
Face property solves the problems, such as that there are technical limitations for collage-credit data acquisition.
Referring to Fig. 3, the first embodiment based on credit-graded approach of the present invention proposes the of credit-graded approach of the present invention
Two embodiments, in the present embodiment, the step S30 includes:
Step S31, it is corresponding according to the scoring of the abnormal reference of each other users, relationship grading and the grading of each relationship
Preset weights establish the mapping relations of other users, abnormal reference scoring and preset weights；
In the present embodiment, on the basis of original first embodiment, the corresponding relationship between wherein data is carried out
It arranges, i.e., can be facilitated by the building of mapping relations and quickly find required abnormal reference scoring in subsequent processing data
And preset weights.Additionally can establish relationship grading and relationship grading in target user abnormal reference grade average with
The relationship is graded the mapping relations between corresponding weight.
The mapping relations data correspondence of foundation is input in the arithmetic unit including formula P=∑ Wi × Qi by step S32,
To export the reference overall score of the associated every other user of the target user, wherein P be target user it is associated it is all its
The reference overall score of his user, Wi are that the abnormal reference of associated i-th of the other users of the target user scores, Qi i-th
The corresponding preset weights of relationship grading of a other users, i ∈ [1, n], n are that the associated other users of the target user are total
Number.
Wherein arithmetic unit can be a part of terminal handler, can also install in third party device.Carrying out data
When processing calculates, it is only necessary to which the data for including in the mapping table established in abovementioned steps input wherein can be obtained other
The abnormal reference overall score of user.When mapping relations data are input to arithmetic unit, the operation language comprising variable can be used
Sentence, mapping relations data, which are brought into, can be completed operation in variable, not repeat excessively herein.The present embodiment passes through mapping relations
And the use of arithmetic unit, helping terminal correspondence when the reference overall score of other users calculates that can quickly find needs makes
Data improve the calculating power of reference scoring, efficient and convenient.
Referring to fig. 4, first embodiment based on credit-graded approach of the present invention proposes the of credit-graded approach of the present invention
Three embodiments, in the present embodiment, the step S10 includes:
Step S11 obtains all kinds of economic activity data of target user, and all kinds of economic activity data is input to
In goal-selling user credit Rating Model, to obtain the initial credit scoring of the target user；
Wherein all kinds of economic activity data may include basic information, and financial asset information communicates IP address information, awarded
Letter refund information, business handling provide data information and other business and follow treaty data, other described business have been followed treaty data packet
Include situation etc. of following treaty of the business such as the business of hiring a car, credit purchase machine.
Goal-selling user credit Rating Model is referred to the existing prior art including machine learning and is configured,
It is only scored and is used by the initial credit that calling model obtains target user herein, detailed code of points and model are adopted
Algorithm is herein without limiting.
Step S12, obtains the corresponding economic activity data of the associated each other users of the target user, and from each
The corresponding abnormal economic activity data of the other users are extracted in the corresponding economic activity data of other users, are obtained with statistics every
The abnormal data amount of a other users；
Wherein, the economic activity data of other users can be compared to the narrow range that target user is related to, such as may include
Basic information, the information of credit refund and communication IP address information, do not include financial asset information and other business are followed treaty number
According to.Wherein abnormal data can be screened to obtain by the economic activity data of each other users of acquisition, then according to different
The number of regular data counts to obtain the abnormal data amount of each other users.
It should be noted that since amount of user data is very big, in order to obtain the reference situation of associated other users not
Need the reference code of points entirely by reference to target user to carry out strict operation, can by abnormal data amount number come slightly
Slightly evaluate the credit situation of other users.Wherein, abnormal data amount, Ke Yishe are obtained from the economic activity data of other users
Set different abnormal data monitoring conditions, it is big that non-repayment amount of e.g. exceeding the time limit is greater than the first preset value, the number that exceeds the time limit not refund
Belong at least one of library third party exception IP in being equal to the second preset value and communicating IP address, if meeting abnormal data
Monitoring condition is it is believed that this economic activity of other users is abnormal, the corresponding economic activity data of this economic activity
It is denoted as abnormal economic activity data.When the economic activity data statistics for other users determines that all abnormal datas monitor item
It is i.e. statistics available after the result of part to obtain abnormal data amount.
It should also be noted that, being monitored due to may relate to multinomial default abnormal data for a certain item economic activity data
Condition, but practical evaluation and test abnormal index consideration is different dimensions, can repeat the statistics of abnormal data amount, alternatively,
The case where a certain item economic activity data can be met with multinomial default abnormal data monitoring condition, to this economic activity number
It is handled again according to remove, guarantees the accuracy of data volume.
Step S13, it is pre- between abnormal data amount and abnormal reference scoring according to the abnormal data amount of each other users
If inquiring the corresponding abnormal reference scoring of each other users in mapping table, wherein abnormal data amount is more, corresponding different
Chang Zhengxin scoring is higher.
In the present embodiment, be by the abnormal data amounts of each other users number to assess obtain other corresponding use
The abnormal reference at family scores, and is preset with the mapping table of abnormal data amount and abnormal reference scoring among these, can will
The particular number of abnormal data amount finds the corresponding abnormal reference of the keyword from mapping table and comments as keyword
Point.Due to being by the corresponding other users of abnormal data situation of other users when obtaining the reference overall score of target user
Reference overall score is as minuend, so if the reference situation of other users is better, minuend should be smaller, i.e., abnormal reference
It scores smaller, abnormal data amount is few；Conversely, the reference situation of other users is poorer, minuend is bigger, i.e., abnormal reference scoring is got over
Greatly, abnormal data amount is more.The present embodiment provides the reference scoring for how obtaining target user and associated other users
Method provides data basis for the smooth expansion of credit scoring.
Further, the default mapping table in the present embodiment between abnormal data amount and abnormal reference scoring can lead to
It crosses data analysis to obtain, such as can be debugged in entire credit investigation system building process according to sample data, firstly for
The abnormal data amount of each different associated other users, or abnormal data amount is referred to each independent abnormal data
Section can correspond to the initial abnormal reference scoring of setting, then correspond to abnormal data amount or the vector in abnormal data section
Value, which is input to, to be exported in neural network as a result, then by the reference difference condition pair of actual sample data and output result
Neural metwork training iteration finally obtains the abnormal data amount for meeting iterated conditional and different to correct initial abnormal reference scoring
The mapping relations of Chang Zhengxin scoring, can measure each user's by the abnormal data of each associated other users with this
Abnormal reference scoring.
Referring to Fig. 5, any embodiment of the first embodiment based on credit-graded approach of the present invention into 3rd embodiment
It is proposed the fourth embodiment of credit-graded approach of the present invention, in the present embodiment, the step S20 includes:
Step S21 obtains the historical sample interaction data between the target user and other users, and passes through clustering algorithm
Clustering is carried out to the historical sample interaction data, to obtain multiple interaction gradings section；
Historical sample interaction data can be a part chosen in all data between target user and other users
Data.The step of wherein carrying out clustering to above-mentioned historical sample interaction data by clustering algorithm includes: by inputting base
Because of the object set of expression matrix building clustering algorithm, the object set is by the corresponding group of data points of all historical sample interaction datas
At；K data point is chosen from the object set, wherein K is greater than or equal to 2, and with the K data in the object set
Point is that cluster centre establishes aggregate of data respectively；Cluster iterative operation is carried out to by the aggregate of data of cluster centre of K data point, with
When aggregate of data after iteration meets preset termination condition, the corresponding number of K cluster centre of newest an iteration is exported
According to cluster, and using K aggregate of data range of output as multiple interaction gradings section.
Wherein, clustering algorithm for example can be K-means algorithm.Cluster iterative operation is being carried out using the aggregate of data of formation
When, each data point in the space map for inputting gene expression matrix is divided again according at a distance from different data cluster
Class, such as data point is adjusted to apart from aggregate of data where nearest cluster centre, it is operated with completing an iteration.In addition, its
In the preset termination condition difference of square distance sum that can be data point in adjacent iterative operation twice to cluster centre be less than
Default error threshold, alternatively, the number of iterative operation reaches preset times threshold value.It should be noted that after completing iteration, most
End form may act as different interaction grading sections at the range of aggregate of data.
Step S22 obtains the interaction data in preset time between the target user and each other users, according to institute
It states interaction data and determines the corresponding affiliated interaction grading section of each other users；
In the present embodiment, such as the interaction for needing to obtain can be determined according to the business application time of different target user
Preset time range corresponding to data, be then referred to aforementioned clustering method by the interaction data in this period into
Row is sorted out, corresponding to obtain interaction grading section belonging to each other users in preset time.
Step S23, by the corresponding preset relation grading of interaction grading section as in affiliated interaction grading section it is each its
The relationship of his user is graded, wherein the corresponding endpoint value in interaction grading section is bigger, preset relation grading is higher.
It should be noted that interaction grading section is corresponding with preset relation grading in advance, wherein each interaction grading section
At least there are two endpoint values, and the endpoint value of corresponding position is bigger, and the grading of relationship belonging to other users is higher.It should be understood that
In this programme it is each interaction grading section be independent from each other, i.e., value therein is not overlapped, each user have and it is only possible directly
It connecing and is divided into a relationship grading, minimum value and maximum value for section are corresponding two endpoint values in section, for
For each other users, if arbitrary value is bigger than the arbitrary value in other certain sections in affiliated section, then it represents that relationship is commented
Grade is higher, furthermore can also distinguish determination according to the size of the endpoint value at interaction grading section both ends, section both ends are corresponding
The endpoint value of position thinks that more greatly preset relation grading is higher.Thus this programme is by the way that by clustering algorithm, help realizes pass
It is the acquisition of grading, has done good differentiation for different user social relationships how far.
In other embodiments, it can also grade to the relationship of determining other users and optimize adjustment.For example, can be with
It is to execute following steps after carrying out classification and obtaining the relationship grading of each other users:
Step S40 obtains the address book data of target terminal user, and determines to whether there is in the address book data and appoint
The corresponding communication number of one other users；
It should be noted that is usually stored in the address list of user terminal is frequent contact information, therefore can incite somebody to action
This is as first order screening criteria, thus according to first order screening criteria by there are communication numbers in target user's address book data
Other users screen.
Step S50 obtains the mesh when communication number corresponding there are any other users in the address book data
There are the second of the correspondence other users terminal of communication number in the first history IP address and address book data of mark user terminal
History IP address, and count identical IP address number in the first history IP address and the second history IP address；
Above-mentioned steps S50 is the second level screening carried out for the other users optimized and revised, be can be target user's end
All history IP address used in holding and by history IP address used in the other users terminal that obtains of level-one screening into
Row registration determines.Under normal circumstances, IP address equally indicates that target user and other users appear in the same local mistake, together
When occur identical IP address number it is more, indicate to contact tightness between target user and the corresponding other users higher.
Step S60, when identical IP address number is greater than preset value, by other corresponding use of the second history IP address
The relationship grading at family promotes a grading, and executes step: obtaining the corresponding preset weights of each relationship grading.
When the identical IP address number that statistics obtains is more than preset value, it is believed that original relationship grading is slightly lower, practical to be somebody's turn to do
The social relationships of other users and target user get close to that degree is higher, the relationship of the other users can be graded and promote level-one.
It should also be noted that, the quantity for optimizing and revising the other users of grading is not limited herein, as long as meeting above-mentioned
Screening criteria, above-mentioned example is only simply illustrated in the case of individually other users are eligible, for multiple
The case where other users, is referred to execute, and this will not be repeated here.
It is possible to further which certain level range is arranged to relationship grading, that is, there is highest grading, if originally other
The grading of user has been highest grading, even if meeting above-mentioned screening criteria, is not also adjusted.I.e. only when identical IP address number
When greater than preset value, judge whether the relationship grading of the corresponding other users of the second history IP address is highest grading；If
It is no, then the relationship grading of the corresponding other users of the second history IP address is promoted into a grading.
Adjustment is optimized for the relationship grading situation of other users by setting what screening criteria in the present embodiment,
Help more precisely to assess target user and other users gets close to degree, to improve the accuracy of target user's reference.
The present invention also proposes that a kind of credit scoring system, the credit scoring system can be server, and can be has
The device end of credit scoring model.Referring to Fig. 6, in one embodiment, the system comprises:
Module 10 is obtained, the initial credit for obtaining target user scores and the associated other users of the target user
Abnormal reference scoring；
Diversity module 20 obtains each other users for being classified to the associated other users of the target user
Relationship grading, and obtain each relationship and grade corresponding preset weights, wherein relationship grading is higher, the corresponding default power obtained
It is worth bigger；
The acquisition module 10 is also used to be scored according to the abnormal reference of each other users, relationship is graded and each
Relationship is graded corresponding preset weights, obtains the reference overall score of the associated every other user of the target user, and by institute
The reference overall score that initial credit scoring subtracts the every other user is stated, the synthesis credit for obtaining the target user is commented
Point.
Optionally, in another embodiment, the acquisition module includes:
Unit is established, for grading according to the scoring of the abnormal reference of each other users, relationship grading and each relationship
Corresponding preset weights establish the mapping relations of other users, abnormal reference scoring and preset weights；
Arithmetic element, for the mapping relations data correspondence of foundation to be input to the operation including formula P=∑ Wi × Qi
In device, to export the reference overall score of the associated every other user of the target user, wherein P is the associated institute of target user
There is the reference overall score of other users, Wi is that the abnormal reference of associated i-th of the other users of the target user scores, and Qi is
The corresponding preset weights of relationship grading of i-th of other users, i ∈ [1, n], n are the associated other users of the target user
Sum.
Optionally, in another embodiment, the acquisition module further include:
Acquiring unit, for obtaining all kinds of economic activity data of target user, and by all kinds of economic activity data
It is input in goal-selling user credit Rating Model, to obtain the initial credit scoring of the target user；
Statistic unit, for obtaining the corresponding economic activity data of the associated each other users of the target user, and
The corresponding abnormal economic activity data of the other users are extracted, from the corresponding economic activity data of each other users with statistics
Obtain the abnormal data amount of each other users；
Query unit scores for the abnormal data amount according to each other users from abnormal data amount and abnormal reference
Between default mapping table in inquire the corresponding abnormal reference scoring of each other users, wherein abnormal data amount is more, right
The abnormal reference scoring answered is higher.
Optionally, in another embodiment, the diversity module includes:
Cluster cell, for obtaining the historical sample interaction data between the target user and other users, and by poly-
Class algorithm carries out clustering to the historical sample interaction data, to obtain multiple interaction gradings section；
Determination unit, for obtaining the interaction data in preset time between the target user and each other users, with
The corresponding affiliated interaction grading section of each other users is determined according to the interaction data；
Setting unit, for the corresponding preset relation grading in grading section will to be interacted as every in affiliated interaction grading section
The relationship of a other users is graded, wherein the corresponding endpoint value in interaction grading section is bigger, preset relation grading is higher.
Optionally, in another embodiment, the cluster cell includes:
Input subelement, for by input gene expression matrix construct clustering algorithm object set, the object set by
The corresponding group of data points of all historical sample interaction datas at；
Subelement is established, for choosing K data point from the object set, wherein K is greater than or equal to 2, and described
Aggregate of data is established respectively using the K data point as cluster centre in object set；
Iteration subelement, for carrying out cluster iterative operation to by the aggregate of data of cluster centre of K data point, repeatedly
When aggregate of data after generation meets preset termination condition, the corresponding aggregate of data of K cluster centre of newest an iteration is exported,
And using K aggregate of data range of output as multiple interaction gradings section.
Optionally, in another embodiment, the system also includes:
Determining module, for obtaining the address book data of target terminal user, and determine in the address book data whether
There are the corresponding communication numbers of any other users；
Statistical module, for obtaining when communication number corresponding there are any other users in the address book data
There are the correspondence other users terminals of communication number in the first history IP address and address book data of the target terminal user
The second history IP address, and count identical IP address number in the first history IP address and the second history IP address；
Hoisting module is used for when identical IP address number is greater than preset value, and the second history IP address is corresponding
The relationship grading of other users promotes a grading, and executes step: obtaining the corresponding preset weights of each relationship grading.
Optionally, in another embodiment, the hoisting module is also used to when identical IP address number is greater than preset value,
Whether the relationship grading for judging the corresponding other users of the second history IP address is highest grading；And work as second history
The relationship of the corresponding other users of IP address is rated highest grading, by the corresponding other users of the second history IP address
Relationship grading promotes a grading.
The present invention also proposes a kind of computer readable storage medium, is stored thereon with computer program.The computer can
Reading storage medium can be the memory 20 in the server of Fig. 1, be also possible to as ROM (Read-Only Memory, it is read-only to deposit
Reservoir)/RAM (Random Access Memory, random access memory), magnetic disk, at least one of CD, the calculating
Machine readable storage medium storing program for executing include some instructions use so that one with processor terminal device (can be mobile phone, computer,
Server or the network equipment etc.) execute method described in each embodiment of the present invention.
It should be noted that, in this document, the terms "include", "comprise" or its any other variant are intended to non-row
His property includes, so that the process, method, article or the server-side that include a series of elements not only include those elements,
It but also including other elements that are not explicitly listed, or further include for this process, method, article or server-side institute
Intrinsic element.In the absence of more restrictions, the element limited by sentence "including a ...", it is not excluded that wrapping
Include in process, method, article or the server-side of the element that there is also other identical elements.
The serial number of the above embodiments of the invention is only for description, does not represent the advantages or disadvantages of the embodiments.
Through the above description of the embodiments, those skilled in the art can be understood that above-described embodiment side
Method can be realized by means of software and necessary general hardware platform, naturally it is also possible to by hardware, but in many cases
The former is more preferably embodiment.
The above is only a preferred embodiment of the present invention, is not intended to limit the scope of the invention, all to utilize this hair
Equivalent structure or equivalent flow shift made by bright specification and accompanying drawing content is applied directly or indirectly in other relevant skills
Art field, is included within the scope of the present invention.
Claims (10)
1. a kind of credit-graded approach, which comprises the following steps:
Obtain the initial credit scoring of target user and the abnormal reference scoring of the associated other users of the target user；
The associated other users of the target user are classified, obtain the relationship grading of each other users, and obtain every
A relationship is graded corresponding preset weights, and wherein relationship grading is higher, and the corresponding preset weights obtained are bigger；
According to the scoring of the abnormal reference of each other users, relationship grading and the corresponding preset weights of each relationship grading, obtain
It takes the reference overall score of the associated every other user of the target user, and initial credit scoring is subtracted described all
The reference overall score of other users obtains the synthesis credit scoring of the target user.
2. credit-graded approach as described in claim 1, which is characterized in that the abnormal reference according to each other users
Scoring, relationship grading and the corresponding preset weights of each relationship grading, obtain the associated every other use of the target user
The step of reference overall score at family includes:
According to the scoring of the abnormal reference of each other users, relationship grading and the corresponding preset weights of each relationship grading, build
Vertical other users, the mapping relations of abnormal reference scoring and preset weights；
The mapping relations data correspondence of foundation is input in the arithmetic unit including formula P=∑ Wi × Qi, to export the mesh
The reference overall score of the every other user of user-association is marked, wherein P is the reference of the associated every other user of target user
Overall score, Wi are that the abnormal reference of associated i-th of the other users of the target user scores, and Qi is i-th of other users
The corresponding preset weights of relationship grading, i ∈ [1, n], n are the associated other users sum of the target user.
3. credit-graded approach as described in claim 1, which is characterized in that the initial credit scoring for obtaining target user
And the associated other users of target user abnormal reference scoring the step of include:
All kinds of economic activity data of target user are obtained, and all kinds of economic activity data are input to goal-selling user
In credit scoring model, to obtain the initial credit scoring of the target user；
The corresponding economic activity data of the associated each other users of the target user are obtained, and corresponding from each other users
Economic activity data in extract the corresponding abnormal economic activity data of the other users, each other users are obtained with statistics
Abnormal data amount；
Default mapping table according to the abnormal data amount of each other users, between abnormal data amount and abnormal reference scoring
Middle to inquire the corresponding abnormal reference scoring of each other users, wherein abnormal data amount is more, and corresponding exception reference scoring is got over
It is high.
4. credit-graded approach as described in claim 1, which is characterized in that described other use associated to the target user
Family is classified, obtain each other users relationship grading the step of include:
The historical sample interaction data between the target user and other users is obtained, and by clustering algorithm to the history sample
This interaction data carries out clustering, to obtain multiple interaction gradings section；
The interaction data in preset time between the target user and each other users is obtained, with true according to the interaction data
Interaction grading section belonging to fixed each other users are corresponding；
Relationship by the corresponding preset relation grading in interaction grading section as each other users in affiliated interaction grading section
Grading, wherein the corresponding endpoint value in interaction grading section is bigger, preset relation grading is higher.
5. credit-graded approach as claimed in claim 4, which is characterized in that it is described by clustering algorithm to the historical sample
Interaction data carries out clustering, to include: the step of obtaining multiple interaction grading sections
The object set of clustering algorithm is constructed by input gene expression matrix, the object set is by all historical sample interaction datas
Corresponding group of data points at；
K data point is chosen from the object set, wherein K is greater than or equal to 2, and with the K number in the object set
Strong point is that cluster centre establishes aggregate of data respectively；
Cluster iterative operation is carried out to by the aggregate of data of cluster centre of K data point, is met with the aggregate of data after iteration pre-
If when termination condition, exporting the corresponding aggregate of data of K cluster centre of newest an iteration, and by K data of output
Cluster range is as multiple interaction gradings section.
6. credit-graded approach as described in any one in claim 1-5, which is characterized in that described to be associated with to the target user
Other users be classified, obtain each other users relationship grading the step of after, further includes:
The address book data of target terminal user is obtained, and determines and whether there is any other users pair in the address book data
The communication number answered；
When communication number corresponding there are any other users in the address book data, the target terminal user is obtained
There are the second history IP address of the correspondence other users terminal of communication number in first history IP address and address book data, and
Count identical IP address number in the first history IP address and the second history IP address；
When identical IP address number is greater than preset value, the relationship of the corresponding other users of the second history IP address is graded
A grading is promoted, and executes step: obtaining the corresponding preset weights of each relationship grading.
7. credit-graded approach as claimed in claim 6, which is characterized in that described when identical IP address number is greater than preset value
When, the step of grading of the relationships of the corresponding other users of the second history IP address is promoted a grading includes:
When identical IP address number is greater than preset value, judge that the relationship of the corresponding other users of the second history IP address is commented
Whether grade is highest grading；
If it is not, the relationship grading of the corresponding other users of the second history IP address is then promoted a grading.
8. a kind of credit scoring system, which is characterized in that the credit scoring system includes:
Module is obtained, the initial credit for obtaining target user scores and the exception of the associated other users of the target user
Reference scoring；
Diversity module obtains the relationship of each other users for being classified to the associated other users of the target user
Grading, and the corresponding preset weights of each relationship grading are obtained, wherein relationship grading is higher, and the corresponding preset weights obtained are got over
Greatly；
The acquisition module is also used to be scored according to the abnormal reference of each other users, relationship grading and each relationship are commented
The corresponding preset weights of grade obtain the reference overall score of the associated every other user of the target user, and will be described initial
Credit scoring subtracts the reference overall score of the every other user, obtains the synthesis credit scoring of the target user.
9. a kind of terminal, which is characterized in that the terminal includes: communication module, memory, processor and is stored in the storage
On device and the computer program that can run on the processor, realized such as when the computer program is executed by the processor
The step of claim 1 to 7 described in any item credit-graded approaches.
10. a kind of computer readable storage medium, which is characterized in that be stored with computer on the computer readable storage medium
Program realizes the credit-graded approach as described in any one of claims 1 to 7 when the computer program is executed by processor
The step of.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910623505.6A CN110517132A (en) | 2019-07-11 | 2019-07-11 | Credit-graded approach, system, terminal and computer readable storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910623505.6A CN110517132A (en) | 2019-07-11 | 2019-07-11 | Credit-graded approach, system, terminal and computer readable storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110517132A true CN110517132A (en) | 2019-11-29 |
Family
ID=68622453
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910623505.6A Pending CN110517132A (en) | 2019-07-11 | 2019-07-11 | Credit-graded approach, system, terminal and computer readable storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110517132A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112085585A (en) * | 2020-08-03 | 2020-12-15 | 北京贝壳时代网络科技有限公司 | Credit risk level assessment method and system |
| CN113888309A (en) * | 2021-10-09 | 2022-01-04 | 支付宝(杭州)信息技术有限公司 | Credit-based data processing method and device |
-
2019
- 2019-07-11 CN CN201910623505.6A patent/CN110517132A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112085585A (en) * | 2020-08-03 | 2020-12-15 | 北京贝壳时代网络科技有限公司 | Credit risk level assessment method and system |
| CN113888309A (en) * | 2021-10-09 | 2022-01-04 | 支付宝(杭州)信息技术有限公司 | Credit-based data processing method and device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107025509B (en) | Decision making system and method based on business model | |
| CN110222880B (en) | Service risk determining method, model training method and data processing method | |
| CN108921686A (en) | A kind of credit-graded approach and device of personal user | |
| CN109934704A (en) | Information recommendation method, device, equipment and storage medium | |
| CN111311030B (en) | User credit risk prediction method and device based on influence factor detection | |
| CN110288350A (en) | User's Value Prediction Methods, device, equipment and storage medium | |
| CN111639690A (en) | Fraud analysis method, system, medium, and apparatus based on relational graph learning | |
| CN110517132A (en) | Credit-graded approach, system, terminal and computer readable storage medium | |
| CN107977855B (en) | Method and device for managing user information | |
| CN109670933A (en) | Identify method, user equipment, storage medium and the device of user role | |
| CN110689427A (en) | Consumption stage default probability model based on survival analysis | |
| CN113205403A (en) | Method and device for calculating enterprise credit level, storage medium and terminal | |
| CN111061704B (en) | Financial analysis report generation method and equipment | |
| CN112860672A (en) | Method and device for determining label weight | |
| CN107844874A (en) | Enterprise operation problem analysis system and its method | |
| CN107608979A (en) | Identify the method and device of the potential knowledge point sought help of user | |
| KR101462858B1 (en) | Methods for competency assessment of corporation for global business | |
| KR20200130767A (en) | Method and device for evaluating whether cryptocurrency is listed on cryptocurrency market using artificial neural network | |
| Yang et al. | Knowledge discovery on customer churn prediction | |
| CN113850669A (en) | User grouping method and device, computer equipment and computer readable storage medium | |
| CN114511022A (en) | Feature screening, behavior recognition model training and abnormal behavior recognition method and device | |
| CN115115322A (en) | Target group identification method, risk assessment method, apparatus, device and medium | |
| CN112598225A (en) | Evaluation index determination method and apparatus, storage medium, and electronic apparatus | |
| CN116484230B (en) | Method for identifying abnormal business data and training method of AI digital person | |
| CN115115449B (en) | Optimized data recommendation method and system for financial supply chain |
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 |