CN110162958A - For calculating the method, apparatus and recording medium of the synthesis credit score of equipment - Google Patents
For calculating the method, apparatus and recording medium of the synthesis credit score of equipment Download PDFInfo
- Publication number
- CN110162958A CN110162958A CN201811212694.XA CN201811212694A CN110162958A CN 110162958 A CN110162958 A CN 110162958A CN 201811212694 A CN201811212694 A CN 201811212694A CN 110162958 A CN110162958 A CN 110162958A
- Authority
- CN
- China
- Prior art keywords
- equipment
- dimension
- credit score
- under
- feature
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 230000015572 biosynthetic process Effects 0.000 title claims abstract description 111
- 238000003786 synthesis reaction Methods 0.000 title claims abstract description 111
- 238000000034 method Methods 0.000 title claims abstract description 75
- 238000012549 training Methods 0.000 claims description 27
- 238000007635 classification algorithm Methods 0.000 claims description 17
- 238000013058 risk prediction model Methods 0.000 claims description 15
- 230000006399 behavior Effects 0.000 claims description 8
- 238000004422 calculation algorithm Methods 0.000 claims description 7
- 238000000605 extraction Methods 0.000 claims description 6
- 238000004590 computer program Methods 0.000 claims description 5
- 230000000644 propagated effect Effects 0.000 description 17
- 230000006870 function Effects 0.000 description 13
- 230000000694 effects Effects 0.000 description 9
- 238000010586 diagram Methods 0.000 description 8
- 230000008569 process Effects 0.000 description 8
- 238000012545 processing Methods 0.000 description 8
- 230000011273 social behavior Effects 0.000 description 7
- 238000004458 analytical method Methods 0.000 description 6
- 230000004069 differentiation Effects 0.000 description 6
- 238000012954 risk control Methods 0.000 description 6
- 230000001419 dependent effect Effects 0.000 description 5
- 238000012417 linear regression Methods 0.000 description 5
- 238000004364 calculation method Methods 0.000 description 4
- 238000003860 storage Methods 0.000 description 4
- 230000008859 change Effects 0.000 description 3
- 235000013399 edible fruits Nutrition 0.000 description 3
- 238000011156 evaluation Methods 0.000 description 3
- 230000001902 propagating effect Effects 0.000 description 3
- 230000001360 synchronised effect Effects 0.000 description 3
- 230000009471 action Effects 0.000 description 2
- 238000004891 communication Methods 0.000 description 2
- 238000013527 convolutional neural network Methods 0.000 description 2
- 238000003066 decision tree Methods 0.000 description 2
- 238000010606 normalization Methods 0.000 description 2
- 230000035515 penetration Effects 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 230000003542 behavioural effect Effects 0.000 description 1
- 238000004140 cleaning Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 238000002372 labelling Methods 0.000 description 1
- 238000007477 logistic regression Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000002844 melting Methods 0.000 description 1
- 230000008018 melting Effects 0.000 description 1
- 238000005065 mining Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000007637 random forest analysis Methods 0.000 description 1
- 230000002123 temporal effect Effects 0.000 description 1
- 238000005303 weighing Methods 0.000 description 1
- 210000002268 wool Anatomy 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F21/00—Security arrangements for protecting computers, components thereof, programs or data against unauthorised activity
- G06F21/30—Authentication, i.e. establishing the identity or authorisation of security principals
- G06F21/44—Program or device authentication
-
- 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
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0635—Risk analysis of enterprise or organisation activities
-
- 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
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
- G06Q10/06393—Score-carding, benchmarking or key performance indicator [KPI] analysis
-
- 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
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/01—Social networking
Landscapes
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Human Resources & Organizations (AREA)
- Strategic Management (AREA)
- Theoretical Computer Science (AREA)
- Economics (AREA)
- Entrepreneurship & Innovation (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- Marketing (AREA)
- Educational Administration (AREA)
- Development Economics (AREA)
- Tourism & Hospitality (AREA)
- Game Theory and Decision Science (AREA)
- Computer Security & Cryptography (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Primary Health Care (AREA)
- Software Systems (AREA)
- General Engineering & Computer Science (AREA)
- Computer Hardware Design (AREA)
- General Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Health & Medical Sciences (AREA)
- Financial Or Insurance-Related Operations Such As Payment And Settlement (AREA)
Abstract
Disclose the method, apparatus and recording medium of the synthesis credit score for calculating equipment.The method for calculating the synthesis credit score of equipment, comprising: obtain the initial data about an equipment；The feature of multiple dimensions is extracted from the initial data, wherein each dimension includes multiple features corresponding with the dimension；Under each dimension, it is based on and the corresponding each feature of the dimension and weight corresponding with each feature, determines the credit score under each dimension；And based on the credit score under weight corresponding with the multiple dimension and each dimension, determine the synthesis credit score of the equipment.
Description
Technical field
This disclosure relates to service security and risk control field.More specifically to the comprehensive letter for calculating equipment
With the method, apparatus and recording medium divided.
Background technique
The identification of existing equipment Risk be mainly based upon device-fingerprint (for example, can unique identification go out the equipment equipment it is special
Sign or unique device identification) malice tracking.In this case it is necessary to which rogue device is known in advance to be had under a scene
Suspicion of doing evil and give mark it is black, be then associated label in other scenes.Such problems is that equipment must be at one very
Black equipment could be labeled as after doing evil to multiple scenes.Also, malice identification is carried out due to only relying upon device-fingerprint, this
Cannot effectively guard against equipment distort, conflict, protocol entry the case where.In addition, equipment only has 0 and 1 label, it may be assumed that instruction is black
The label of equipment or white equipment.However, can not achieve the differentiation of confidence level for white equipment.To for practical risk
The usage scenario of control is limited.
Summary of the invention
In view of above situation, it is intended to provide the credit score calculation method and device of the differentiation that can be realized confidence level.
According to one aspect of the disclosure, a kind of method for calculating the synthesis credit score of equipment is provided, comprising: obtain
Take the initial data about an equipment；The feature of multiple dimensions is extracted from the initial data, wherein each dimension includes
Multiple features corresponding with the dimension；Under each dimension, based on each feature corresponding with the dimension and with each feature
Corresponding weight determines the credit score under each dimension；And it is based on weight corresponding with the multiple dimension and each dimension
The lower credit score of degree, determines the synthesis credit score of the equipment, wherein the multiple dimension include social activity dimension, security dimension,
At least two in value relevance and primary attribute dimension.
In addition, method according to an embodiment of the present disclosure may further include: by using supervised classification algorithm, being based on
First sample library is trained, and is determined in a manner of by learning automatically corresponding with the dimension each under each described dimension
The weight of feature.
In addition, method according to an embodiment of the present disclosure may further include: by using supervised classification algorithm, being based on
First sample library is trained, and the weight of each dimension is determined in a manner of by learning automatically.
In addition, method according to an embodiment of the present disclosure may further include: by using supervised classification algorithm, being based on
Second sample database is trained to obtain an equipment Risk prediction model；And it is based on the equipment Risk prediction model, it determines
The risk probability of the equipment.
In addition, method according to an embodiment of the present disclosure may further include: the risk probability based on the equipment is adjusted
The synthesis credit score of the whole equipment.
In addition, based on the risk probability of the equipment, adjusting the synthesis in method according to an embodiment of the present disclosure
If the step of credit score, includes: the comprehensive credit score less than first threshold and the risk probability is less than second threshold,
Improve the comprehensive credit score；And if the comprehensive credit score is greater than third threshold value and the risk probability is greater than the 4th threshold
Value, then reduce the comprehensive credit score.
In addition, method according to an embodiment of the present disclosure may further include: when the synthesis credit score of the equipment is full
When sufficient predetermined condition, the synthesis credit score based on the equipment, the determining synthesis credit with the associated other equipment of the equipment
Point.
In addition, being existed in method according to an embodiment of the present disclosure under social dimension based on account associated by equipment
Multiple features and corresponding weight under social scene determine credit score；Under the security dimension, based on doing evil with equipment
The relevant multiple features of behavior and corresponding weight determine credit score；Under value relevance, the value based on equipment itself
Attributive character and corresponding weight determine credit score；And under primary attribute dimension, technical data based on equipment is special
Sign and corresponding weight determine credit score.
According to another aspect of the present invention, it provides a kind of for calculating the device of the synthesis credit score of equipment, comprising: obtain
Unit is taken, for obtaining the initial data about an equipment；Feature extraction unit, it is multiple for being extracted from the initial data
The feature of dimension, wherein each dimension includes multiple features corresponding with the dimension；And comprehensive credit score computing unit, it uses
In under each dimension, it is based on and the corresponding each feature of the dimension and weight corresponding with each feature, determines each
Credit score under dimension, and based on the credit score under weight corresponding with the multiple dimension and each dimension, determine described in set
Standby synthesis credit score, wherein the multiple dimension includes in social dimension, security dimension, value relevance and primary attribute dimension
At least two.
In addition, the comprehensive credit score computing unit further comprises in device according to an embodiment of the present disclosure: instruction
Practice unit, it is true in a manner of by learning automatically for being trained based on first sample library by using supervised classification algorithm
The weight of each feature corresponding with the dimension under each fixed described dimension.
In addition, the comprehensive credit score computing unit further comprises in device according to an embodiment of the present disclosure: instruction
Practice unit, it is true in a manner of by learning automatically for being trained based on first sample library by using supervised classification algorithm
The weight of fixed each dimension.
In addition, device according to an embodiment of the present disclosure may further include: risk probability computing unit, for passing through
Using supervised classification algorithm, it is trained based on the second sample database to obtain an equipment Risk prediction model, and set based on described
Standby risk forecast model, determines the risk probability of the equipment.
In addition, device according to an embodiment of the present disclosure may further include: adjustment unit, for being based on the risk
The risk probability for the equipment that probability calculation unit determines adjusts the equipment that the comprehensive credit score computing unit determines
Synthesis credit score.
In addition, the adjustment unit is configured in device according to an embodiment of the present disclosure: if described
Comprehensive credit score is less than first threshold and the risk probability is less than second threshold, then improves the comprehensive credit score；And such as
Comprehensive credit score is greater than third threshold value described in fruit and the risk probability is greater than the 4th threshold value, then reduces the comprehensive credit score.
In addition, device according to an embodiment of the present disclosure may further include: association propagation unit, for being set when described
When standby synthesis credit score meets predetermined condition, the synthesis credit score based on the equipment, it is determining with the equipment it is associated its
The synthesis credit score of his equipment.
In addition, the comprehensive credit score computing unit is further configured in device according to an embodiment of the present disclosure
Are as follows: under social dimension, based on multiple features of the account associated by equipment under social scene and corresponding weight come really
Determine credit score；Under security dimension, determined based on multiple features relevant to the behavior of doing evil of equipment and corresponding weight
Credit score；Under value relevance, credit score is determined based on the property of value feature of equipment itself and corresponding weight；And
Under primary attribute dimension, technical data feature based on equipment and corresponding weight determine credit score.
In accordance with a further aspect of the present invention, it provides a kind of for calculating the device of the synthesis credit score of equipment, comprising: deposit
Reservoir, for storing computer program；And processor, for when execution computer program stored in the storage unit
When, perform the steps of the initial data obtained about an equipment；The feature of multiple dimensions is extracted from the initial data,
Wherein each dimension includes multiple features corresponding with the dimension；Under each dimension, based on corresponding with the dimension each
A feature and weight corresponding with each feature, determine the credit score under each dimension；And it is based on and the multiple dimension
Credit score under corresponding weight and each dimension determines the synthesis credit score of the equipment, wherein the multiple dimension includes
At least two in social dimension, security dimension, value relevance and primary attribute dimension.
According to another aspect of the invention, a kind of computer readable recording medium is provided, is calculated for storage on it
Machine program performs the steps of the initial data obtained about an equipment when executing the computer program by processor；
The feature of multiple dimensions is extracted from the initial data, wherein each dimension includes multiple features corresponding with the dimension；
Under each dimension, it is based on and the corresponding each feature of the dimension and weight corresponding with each feature, determines each dimension
Credit score under degree；And based on the credit score under weight corresponding with the multiple dimension and each dimension, determine described in set
Standby synthesis credit score, wherein the multiple dimension includes in social dimension, security dimension, value relevance and primary attribute dimension
At least two.
In the method and apparatus of synthesis credit score according to an embodiment of the present disclosure for calculating equipment, one is realized
Kind is in various dimensions to the model of the credit rating of equipment.Black equipment is marked with single device-fingerprint is based only upon in the prior art
Scheme is compared, can more effectively cope with equipment distort, conflict, protocol entry the case where.In addition, passing through the reality according to the disclosure
The method and apparatus for applying example can calculate the synthesis credit score of equipment.With in the prior art only by 0 and 1 come marking arrangement
Scheme compare, the differentiation of confidence level can be further realized.
Detailed description of the invention
Fig. 1 is to show the schematic diagram of the application environment of embodiment of the disclosure；
Fig. 2 is to show the process of the method for synthesis credit score according to an embodiment of the present disclosure, for calculating equipment
Figure；
Fig. 3 be show according to another embodiment of the present disclosure, the method for synthesis credit score for calculating equipment
Flow chart；
Fig. 4 is to show the method for synthesis credit score according to the another embodiment of the disclosure, for calculating equipment
Flow chart；
Fig. 5 is the schematic block diagram of the data flow in the method shown according to the embodiment of the present disclosure；
Fig. 6 shows K-S of the synthesis credit score obtained by the method according to the embodiment of the present disclosure under real scene
The evaluation and test effect data curve graph of value；
Fig. 7 is to show the function of the device of synthesis credit score according to an embodiment of the present disclosure, for calculating equipment
Property block diagram；
Fig. 8 be show according to another embodiment of the present disclosure, the device of synthesis credit score for calculating equipment
Functional block diagram；
Fig. 9 is to show the device of synthesis credit score according to the another embodiment of the disclosure, for calculating equipment
Functional block diagram；
Figure 10 shows one as hardware entities according to the device of the synthesis credit score for calculating equipment of the disclosure
A example；And
Figure 11 shows the schematic diagram of computer readable recording medium according to an embodiment of the present disclosure.
Specific embodiment
Each preferred embodiment of the disclosure is described below with reference to accompanying drawings.It provides referring to the drawings
Description, to help the understanding to the example embodiment of the disclosure as defined by appended claims and their equivalents.It includes side
The various details of assistant's solution, but they can only be counted as illustratively.Therefore, it would be recognized by those skilled in the art that
Embodiment described herein can be made various changes and modifications, without departing from the scope and spirit of the disclosure.Moreover, in order to
Keep specification more clear succinct, by omission pair it is well known that the detailed description of function and construction.
Firstly, briefly describing the application environment of embodiment of the disclosure.As shown in Figure 1, server 10 is connected by network 30
It is connected to multiple equipment 20.The multiple equipment 20 can be the equipment for actually carrying out various businesses.Although by equipment 20 in Fig. 1
It is uniformly shown as mobile phone, but the disclosure is not limited to that.It will be understood to those skilled in the art that equipment 20 can also be
The equipment of any other type, such as PDA (personal digital assistant), tablet computer, desktop computer.The server 10 can
To be the device of the synthesis credit score for calculating equipment described below.The network 30 can be it is any kind of wired or
Wireless network, such as internet.It should be appreciated that the quantity of server 10 shown in FIG. 1 and equipment 20 be it is schematical, without
It is restrictive.
Next, by referring to Fig. 2 description according to a kind of embodiment of the disclosure, the synthesis credit score that is used to calculate equipment
Method.The method server 10 shown in Fig. 1 executes.As shown in Fig. 2, the method includes following steps
Suddenly.
Firstly, obtaining the initial data about an equipment in step S201.
In application environment shown in Fig. 1, server 10 is connected and communicated with multiple equipment.Multiple equipment is in its operation
The various initial data of itself (for example, Social behaviors data, device attribute data, application data etc.) is sent in the process
To server 10.For example, Social behaviors data may include logging in log, login times within a predetermined period of time, account
Log in the data such as number of days.Device attribute data may include stem version data, basic software and hardware data, using zone data
Deng.Application data may include application version data, operation data in the application etc..To service
In device 10, it is stored with the mass data from multiple equipment 20.In the present specification, with the synthesis credit score of one equipment of calculating
For be described.Therefore, it is necessary to obtain first from the mass data stored in server and to calculate comprehensive credit score
The relevant initial data of equipment.
Next, extracting the feature of multiple dimensions from the initial data in step S202.
For example, the multiple dimension may include in social dimension, security dimension, value relevance and primary attribute dimension
At least two.Wherein, each dimension includes multiple features corresponding with the dimension.Due to Social behaviors be networked devices must
Indispensable behavior, and black equipment is frequently utilized that Social behaviors to be attacked, therefore will feature relevant to Social behaviors
The feature being individually classified as under social dimension is evaluated in order to more accurately be analyzed for the feature under social dimension
Confidence level of the equipment under social dimension.For example, account associated by major integration excavating equipment is in social activity under social dimension
Various actions attribute under scene, thus, for example, can be by log-on message, registration information, liveness information (for example, pre- timing
Between number of operations etc. in section), message behavioural information (for example, send message to good friend's equipment or receive message from good friend's equipment)
Deng as feature corresponding with social dimension.In addition, different from social dimension, security dimension is primarily directed to the behavior of doing evil.
For example, the feature under security dimension may further include by account associated by equipment more scenes do evil information into
Multiple features obtained from row statistics.Although equally may relate to Social behaviors under security dimension, under security dimension
The not only social scene for being concerned with information of doing evil, and paying close attention under security dimension, further includes the work under other scenes
Dislike information.In addition, the component that the feature under security dimension can further include equipment distorts suspicious degree, whether is equipped with mould
The features such as quasi- device (ulling up wool activity for all kinds of).Feature relevant to the behavior of doing evil individually is classified as under security dimension
Feature is analyzed in order to more accurately for the feature under security dimension credible under security dimension come valuator device
Degree.Value relevance may include property of value feature (e.g., the Time To Market of device model, the equipment of the model of equipment itself
Deng), the payment data that can also further integrate account associated by social liveness information or equipment is tieed up as with value
Spend corresponding feature.That is, there may be the overlappings of Partial Feature between multiple dimensions.Primary attribute dimension mainly for
The technical data of equipment, for example, can by system version information, underlying hardware information, software information, made using regional information etc.
For feature corresponding with primary attribute dimension.
Certainly, the quantity of the dimension is not limited in above example, can also constantly be expanded according to different scenes,
And the specific name of dimension can also be adjusted and be changed according to different scenes.
In addition, the step of extracting the feature of multiple dimensions from the initial data may further include following steps.
Initial data is pre-processed first.Specifically, it is described pretreatment may include data it is isometric filling and it is different
Regular data cleaning etc..Then, for executing normalized by pretreated data.For example, following formula can be passed through
(1) or (2) Lai Zhihang normalized.
F (x)=(a+x)/(b+x) (1)
Wherein, F (x) indicates the feature obtained after normalized, and x indicates initial data.A and b respectively indicates normalization
Parameter, can according to different data fields index be adjusted.
For example, by the way that the initial data x of the login number of days as account is input to above formula, between output 0~1
Value, as the feature obtained after normalized.Certainly, for different initial data can using different functions come into
Row normalization, and method for normalizing is also not necessarily limited to enumerated above two kinds.
In addition, as a kind of possible embodiment, for the feature obtained after normalized, can also further tie
Time change and liveness function are closed, history feature weighting is carried out.If the duration of a feature is long or liveness is high,
So this feature will be given biggish weight.Specific formula is as follows:
T (x)=[cf (ti-t0)+(1-c) f (tA)] F (x) (3)
Wherein, T (x) indicates the feature obtained after history feature weighting, and f (ti-t0) indicates the time change function of feature,
Wherein t0 indicates that feature just starts the time occurred, and ti indicates current time existing for feature, and f (tA) indicates the liveness of feature
Function, c indicate a setting coefficient and can be arranged according to different scenes.Alternatively, of course, the time can also be based only upon
One in variation function and liveness function carries out characteristic weighing.
Then, corresponding with the dimension each based on being extracted in step S202 under each dimension in step S203
A feature and weight corresponding with each feature, determine the credit score under each dimension.
Here, under each dimension, there are corresponding basic score models.Basic score mould under every dimension
Type includes weight corresponding with multiple features under the dimension etc..Weight corresponding with each feature is predetermined.For example,
The basic score model can be trained based on first sample library by supervised classification algorithm, it is described each with determination
The weight of each feature corresponding with the dimension under a dimension.First sample library may include: general black and white library, registration black and white
Library, malice number library, the black library of plusing good friend etc..For example, general black and white library may include under all scenes known rogue device and
Feature under each dimension of normal device.Registration black and white library may include the known malice marked by registering operation
Feature under each dimension of equipment and normal device.Malice number library may include account corresponding to known rogue device
Information or equipment identification information.The black library of plusing good friend may include operating the known rogue device marked by plusing good friend
Feature under each dimension.Furthermore, it is possible to integrally regard basic score model and first sample library as a modeling engine.Supervision
Sorting algorithm is including a target variable (dependent variable, i.e., credit score corresponding with the feature under the dimension) and for predicting target
The predictive variable (independent variable, i.e., feature corresponding with the dimension) of variable.The basic score mould can be built by these variables
Type, hence for a known predictive variable value, available corresponding target variable value.This model of repetition training, directly
Scheduled accuracy can be reached on training dataset to it.Specifically, in embodiment of the disclosure, sample database includes
It is knowing, as input independent variable certain dimension under feature and it is known, as output dependent variable credit score.Training is base
In known independent variable and dependent variable training weight.By the training learning process of algorithm, weight is continuously adjusted, can finally be looked for
The weight of correct credit score is obtained to one group.Once weight determines, model has also been determined that.Then model is used for sample database
Characteristic variable in addition, and then obtain corresponding credit score.For example, the algorithm of supervised classification include: linear regression, decision tree,
Random forest, K nearest neighbor algorithm, logistic regression etc..
For operand is small and the good consideration of interpretation, for example, in one embodiment, it can be by using linear
Homing method is trained to be based on first sample library.For example, with SiIndicate the credit score of the equipment under i-th of dimension.It so can be with
S is calculated by following formula (4)i:
Wherein n is the feature sum under i-th of dimension, Ti(xj) it is j-th of characteristic variable corresponding with i-th of dimension,
dijFor weight corresponding with j-th of characteristic variable.After the completion of based on the training in first sample library, obtain and i-th of dimension
The weight d of corresponding all featuresi1~din.Then, based on the d determinedi1~din, by inputting setting for current expectation calculating
Standby characteristic variable corresponding with i-th of dimension, to calculate the credit score of the equipment under i-th of dimension.Each height dimension
Degree can all calculate the credit score of such as 0-100.
Specifically, based on the feature under social dimension, such as based on account associated by equipment under social scene
Multiple features and corresponding weight, can determine the credit score under social dimension.In other words, by will be under social dimension
Feature be input to it is corresponding with social dimension, according to the above method basic score model that training obtains in advance, can
Using exporting the basic score model as the credit score under social dimension.Credit score under social dimension is that reflection equipment exists
The score of credibility under social dimension.Since basic score model corresponding with social dimension is based under social dimension
Feature training obtains, therefore being capable of more accurately credibility of the assessment equipment under social dimension.
Based on the feature under security dimension, such as based on multiple features relevant to the behavior of doing evil of equipment and correspondence
Weight, can determine the credit score under security dimension.In other words, it is tieed up by being input to the feature under security dimension with safety
Spend it is corresponding, according to the above method obtained basic score model of training in advance, can be by the basic score model
Output as the credit score under security dimension.Credit score under security dimension is the credible journey for reflecting equipment under security dimension
The score of degree.Since basic score model corresponding with security dimension is obtained based on the feature training under security dimension, because
This being capable of more accurately credibility of the assessment equipment under security dimension.
Property of value feature and corresponding weight based on the feature under value relevance, such as based on equipment itself,
It can determine the credit score under value relevance.In other words, corresponding with value relevance by being input to the feature under value relevance
, according to the above method obtained basic score model of training in advance, can be by the output of the basic score model
As the credit score under value relevance.Credit score under value relevance is point for reflecting credibility of the equipment under value relevance
Number.It, can since basic score model corresponding with value relevance is obtained based on the feature training under value relevance
More accurately credibility of the assessment equipment under value relevance.
Based on the feature under primary attribute dimension, such as technical data feature based on equipment and corresponding weight,
It can determine the credit score under primary attribute dimension.In other words, by the way that the feature under primary attribute dimension to be input to and basis
Attribute dimensions are corresponding, train obtained basic score model in advance according to the above method, which can be commented
The output of sub-model is as the credit score under basic attribute dimensions.Credit score under primary attribute dimension is reflection equipment on basis
The score of credibility under attribute dimensions.Since basic score model corresponding with primary attribute dimension is based on primary attribute
Feature training under dimension obtains, therefore being capable of more accurately credibility of the assessment equipment under primary attribute dimension.
Finally, based on the credit score under weight corresponding with the multiple dimension and each dimension, being determined in step S204
The synthesis credit score of the equipment.
Here, weight corresponding with each dimension is also predetermined.For example, with the above, determining and every
The mode of the weight of the corresponding each feature of one dimension similarly, can be based on the first sample by using supervised classification algorithm
This library is trained, to determine the weight of each dimension.Similarly, small for operand and interpretation is good examines
Consider, for example, in one embodiment, the training of first sample library can be based on by using linear regression method.It can incite somebody to action
The model that training obtains in this way is known as multidimensional scorecard model.For example, indicating the synthesis credit score of equipment with G.Can so it lead to
Following formula (5) is crossed to calculate G:
Wherein m is dimension sum, SiFor the credit score of the equipment under i-th of dimension, eiIt is corresponding with i-th of dimension
Weight.After the completion of based on the training in first sample library, weight e corresponding with all dimensions is obtained1~em.Then, based on determination
Good e1~em, by inputting the credit score corresponding with all dimensions determined in step S203, to calculate the synthesis credit of equipment
Point.
Depending on different application scenarios, weight corresponding with each dimension is different.For example, in social application scene
In, social dimension is more important, therefore weight corresponding to social dimension is bigger.For another example, in financial application scene, safety
Dimension is more important, therefore weight corresponding to security dimension is bigger.
As it can be seen that realizing one kind by method according to an embodiment of the present disclosure and commenting in various dimensions equipment progress credit
The model of grade.Compared with being based only upon single device-fingerprint in the prior art and marking the scheme of black equipment, can more effectively it answer
Equipment is distorted, is conflicted, protocol entry the case where.In addition, by method according to an embodiment of the present disclosure, it can be based on social activity
Dimension, security dimension, value relevance, primary attribute dimension calculate the synthesis credit score of equipment.Due to the feature to acquisition into
One step is classified as the feature under different dimensions, and training basic score model corresponding with different dimensions, therefore can be more acurrate
The synthesis credit score of ground calculating equipment.Also, in the prior art only by 0 and 1 come the scheme of marking arrangement compared with, Neng Goujin
The differentiation of one step realization confidence level.
The initial data for being supplied to server is the number of devices of the magnanimity updated every predetermined period (for example, daily)
According to, social attribute data etc., therefore, the synthesis credit score of equipment also can dynamically update calculating, to cope with changing for device attribute
Become.
It is pointed out here that first sample library includes the sample of all scenes.In other words, first sample library can be with
Regard general sample database as.It is the universal model being applicable in for all scenes based on the model that first sample library trains.However, being
Better adapt to various special scenes, it is also desirable to can on the basis of universal model further combined with special purpose model, to mention
Accuracy and discrimination under high various special scenes.
Therefore, as alternatively possible embodiment, after calculating the synthesis credit score of equipment, according to the disclosure
Method can further include adjustment equipment synthesis credit score the step of.
Specifically, Fig. 3 show according to another embodiment of the present disclosure, synthesis credit score for calculating equipment
Method.As shown in figure 3, the method includes the steps S201~S204, this and the complete phase of processing described above in reference to Fig. 2
Together.Therefore, for the sake of in order to avoid redundancy, its details is repeated no more here.Also, in addition to the above step S201~
Except S204, the method further includes following steps.
In step S301, by using supervised classification algorithm, it is trained based on the second sample database to obtain an equipment wind
Dangerous prediction model.
It is pointed out here that the second sample database can be identical as first sample library.Alternatively, alternatively, the second sample
This library is also possible to a subset in first sample library.For example, the second sample database is the sample database for special scenes.For example,
Under the specific application scene that registration intercepts, the second sample database can be set to registration black and white library.Specifically, in the disclosure
Embodiment in, sample database include it is known, as input independent variable multidimensional characteristic and it is known, as output dependent variable
Risk probability.For example, equipment Risk prediction model includes function corresponding with multidimensional characteristic and parameter etc..Training is based on
Independent variable and dependent variable those functions of training and parameter known.By the training learning process of algorithm, continuously adjust function and
Parameter can finally find one group and obtain the function and parameter of correct risk probability.Once function and parameter determine, also determine that
Model.Then model is used for the characteristic variable other than sample database, and then obtains corresponding risk probability.
For example, the second sample database can be trained using method identical with the above linear regression,
Or it can also be using other supervised classification algorithms different from the above linear regression method come to the second sample database
It is trained.For example, second can be based on by nonlinear model, such as decision tree (GBDT), convolutional neural networks (CNN)
Sample database is trained.
Then, in step S302, it is based on the trained equipment Risk prediction model, determines that the risk of the equipment is general
Rate.For example, a part of feature inputted in multiple dimensions of the step S202 equipment extracted or whole spies can be passed through
Levy the risk probability to determine the equipment.
Finally, based on the risk probability of the equipment, adjusting the synthesis credit score of the equipment in step S303.
Therefore, the model trained based on the second sample database is the special purpose model for special scenes.According to the disclosure
The method for calculating the synthesis credit score of equipment can be applied to the risk control system of multiple business, and including but not limited to activity is anti-
Brush, malicious registration, identity authentication, stolen protection etc..In addition, although the disclosure is controlled using business risk as point of penetration, tool
The application field of the equipment complex credit score of body not only includes security risk control field, and it is more to apply also for recommendation, advertisement etc.
A field.That is, in the biggish different application scene of otherness, by using the die for special purpose for being directed to the application scenarios
Type can effectively improve the accuracy of final calculated comprehensive credit score.
As a kind of possible embodiment, can believe only for high risk and the comprehensive of the equipment in high credible two sections
It is adjusted with point.For example, the risk probability based on the equipment, the step of adjusting the comprehensive credit score may include: as
Comprehensive credit score is less than first threshold described in fruit and the risk probability is less than second threshold, then improves the comprehensive credit score；
And if the comprehensive credit score is greater than third threshold value and the risk probability is greater than the 4th threshold value, reduce the comprehensive letter
With point.
Specifically, it can be adjusted by way of following formula (6).
Wherein, F' is the synthesis credit score of equipment adjusted, and F is the synthesis credit score of the equipment before adjustment, and e is to set
Standby risk probability.Low equipment complex credit score threshold value Se corresponds to the above first threshold, and high equipment complex is believed
With dividing threshold value Sg to correspond to the above second threshold, the two can be made a concrete analysis of to obtain according to business scenario.And equipment
Risk probability threshold value E corresponds to the above third threshold value and the 4th threshold value, can take intermediate probability value 0.5.As it can be seen that
In formula (6), third threshold value and the 4th threshold value are disposed as E.It alternatively, of course, can also be by third threshold value and the 4th
Threshold value is set as different values.The value of weight D can be according to K-S evaluation (discrimination of the normal rogue device) knot in specifically used
The study of fruit inverse iteration obtains.Specifically, the value of D is respectively set with pre- fixed step size in a range intervals, and determines phase
The K-S value answered (related illustrating for K-S value will be described below).Then, corresponding D value conduct when taking K-S value maximum
D value in formula (6).
Therefore, in this embodiment according to the disclosure, by using the calculated equipment based on special purpose model
Risk probability adjusts the synthesis credit score of equipment, can further increase the accuracy of the synthesis credit score of equipment, more effectively
Distinguish black equipment and white equipment in ground.
Furthermore, it is possible to it is very few in the presence of the initial data about a certain equipment, and then the characteristic under each dimension extracted
Very few situation.In such circumstances it is desirable to which the credit score accuracy in computation of this equipment can be further increased.
It therefore, can be by establishing the network with the associated other equipment of equipment as another possible embodiment
Model (that is, facility network described below), the association for then carrying out credit score calculate.It is adopted by the synthesis credit score to equipment
Analysis is propagated with association, is able to ascend the accuracy rate of the synthesis credit score of the infull equipment of data characteristics dimension.Comprehensive credit score
Association to propagate analysis be based on " being likely to be black equipment with the associated equipment of black equipment, with the associated equipment of white equipment
Be likely to be white equipment " hypothesis and carry out.
For example, being associated between equipment and equipment can be associated with by the account logged in equipment.Specifically, in A
The account that user C has been logged in equipment has logged in the account of user D, and the account of the account of user C and user D on B device
Number be friend relation, then can using B device as with the associated equipment of A equipment.Certainly, the interrelational form between equipment is not
It is only limitted to this.It will be understood to those skilled in the art that any other interrelational form can also be suitably applied to the disclosure.
For example, it is also possible to carry out the association of equipment room by data such as the business of login IP/WIFI or clicking operation.It is specific next
It says, can will regard associate device as in same IP sections of equipment, or the equipment in same WIFI network can be regarded as
Associate device, or the equipment for clicking identical link can will be performed as associate device.
It is pointed out here that the association that can not carry out comprehensive credit score to all equipment is propagated.It is being closed
Before connection is propagated, the equipment as propagating source can be judged.Specifically, Fig. 4 is shown according to the another of the disclosure
The flow chart of the method for embodiment.As shown in figure 4, the method includes the steps S201~S204, and determined in step S204
Out after the synthesis credit score of equipment, further comprise the steps.
In step S401, judge whether the synthesis credit score of the equipment meets predetermined condition.Such as, it can be determined that it is described
Whether the synthesis credit score of equipment is in specific sections.For example, passing through the range of the calculated comprehensive credit score of step S204
For 0-800.Wherein, if score is closer to two endpoint values (that is, 0,800), the discrimination of equipment is higher, if score
Closer to median (that is, 400), then the discrimination of equipment is lower.It is known as the high equipment of discrimination to firmly believe equipment, it can
The equipment for distinguishing black equipment or white equipment.As an example, will comprehensive credit score less than a threshold value (e.g., 200) and
The equipment that comprehensive credit score is greater than another threshold value (e.g., 600), which is regarded as, firmly believes equipment, and the only score progress to equipment is firmly believed
Association is propagated.
If be judged as YES in step S401, it may be assumed that the synthesis credit score of the equipment meets predetermined condition, then handles progress
To step S402.In step S402, the synthesis credit score based on the equipment is determining and the associated other equipment of the equipment
Comprehensive credit score.For example, can be by the synthesis credit score of the equipment multiplied by scheduled coefficient (for example, coefficient between 0~1)
After propagate to associate device.
Certainly, the credit score of itself can not only be propagated to other associate devices by current device, but also can be received
It is propagated to the credit score from other associate devices.In such a case, it is possible to by the synthesis credit of calculated current device
Divide and integrates with the credit score propagated from other associate devices to obtain final synthesis credit score.For example, can be by that will calculate
The synthesis credit score of current device out, which is input to a weighted model with the synthesis credit score propagated from other associate devices, to be come
To final synthesis credit score.For example, can determine propagated credit based on the correlation degree between two associate devices
The weight divided.
At the association propagation for carrying out comprehensive credit score after describing step S204 shown in Fig. 2 above by reference to Fig. 4
Reason.Certainly, the disclosure is not limited to that.For example, it is also possible to carry out comprehensive credit score after step S303 shown in Fig. 3
It is associated with dissemination process.
Fig. 5 is the schematic frame of the data flow in the synthesis credit score calculation method shown according to the embodiment of the present disclosure
Figure.As shown in figure 5, the initial data including such as Social behaviors data, device attribute data, application data etc is defeated
Enter to modeling engine.In modeling engine, including basic score model and sample database.Alternatively, alternatively, the modeling engine
It can further include facility network.
As mentioned above it is possible, basic score model is basic score model corresponding with multiple dimensions respectively.By being based on
Sample database the basic score model under each dimension is trained and the weight corresponding with each feature that learns with
And the characteristic variable of input, the credit score under each dimension can be obtained.By melting to the credit score under each dimension
It closes, such as by way of calculating comprehensive credit score above in reference to formula (5), available multidimensional scorecard model.Only base
Comprehensive credit score is obtained in the multidimensional scorecard model to correspond to as above in reference to the embodiment described in Fig. 2, and due to instructing
Practice multidimensional scorecard model based on sample database be sample database under generic scenario, thus it is general to can consider that it corresponds to one
Model.
In addition, being based on sample database, the equipment Risk prediction model described above in reference to Fig. 3 can also be obtained.In Fig. 5
In, and first sample library and the second sample database mentioned hereinabove are not differentiated between, but it is collectively shown as sample database.As above
It is described in text, for train multidimensional scorecard model sample database can with for training the sample database of equipment Risk prediction model
It is identical, it can also be different.For example, can be general sample database for training the sample database of multidimensional scorecard model, and it be used to instruct
The sample database for practicing equipment Risk prediction model can be the sample database for dedicated scene, can be the general sample database
A part.
Also, the facility network established and being associated between devices, available facility network correlation model.Example
Such as, above with reference to Fig. 4 description step S402 in, determined based on the facility network correlation model with equipment it is associated other
The synthesis credit score of equipment.
As multidimensional scorecard model and equipment Risk prediction model can be based on above in reference to the embodiment described in Fig. 3
Comprehensive credit score is obtained, and as multidimensional scorecard model and facility network can be based on above in reference to the embodiment described in Fig. 4
Correlation model obtains comprehensive credit score.It certainly, as mentioned above it is possible, can also be pre- based on multidimensional scorecard model, equipment Risk
It surveys model and facility network correlation model obtains comprehensive credit score.In the above embodiment, these models can be carried out
Fusion, and due to having merged the model obtained based on the training of dedicated sample database, so it is considered that it corresponds to a special purpose model.
In Fig. 5, non-essential optimization model is shown with dotted line frame.
After model foundation, need to assess the effect of model.In disaggregated model assessment, one kind is commonly commented
Estimating standard is K-S value.Fig. 6 is shown through the synthesis credit score that obtains according to the method for the embodiment of the present invention under real scene
K-S value evaluation and test effect data curve graph.In Fig. 6, horizontal axis indicate equipment synthesis credit by stages (such as < 100, <
200 ...), the longitudinal axis indicates the accounting of rogue device and total rogue device in the synthesis credit by stages, or is in
The accounting of the normal device of the synthesis credit by stages and total normal device.As shown in fig. 6, illustrating three curves, respectively
For the difference curve of curve corresponding to curve corresponding to rogue device, normal device and the two curves.Also, take difference
Be worth the maximum value of curve, it may be assumed that largest interval between curve corresponding to curve corresponding to rogue device and normal device away from
From being exactly K-S value.From fig. 6 it can be seen that K-S > 64%, therefore model has relatively good forecasting accuracy, thus according to this
The method of open embodiment can efficiently differentiate out rogue device and normal device.
Primarily directed to the credit scoring of user identity, there is no actually available for credit scoring in currently available technology
For the credit rating of equipment.Such problems is, with the fast development of internet and Internet of Things, the quantity of networked devices
The quantity of user is alreadyd exceed, if black production is attacked using multiple devices, has no idea to efficiently identify black equipment.From
And the accuracy for improving the detection for being directed to device security is a urgent problem to be solved.Due in the reality according to the disclosure
Applying calculate in the method for example is equipment rather than the synthesis credit score of user identity, therefore other business sides do not need to know use
The details at family, do not need customer attribute information to be synchronized to third party yet and provide for credit level based on equipment and sentence
Not, the use of privacy of user data is avoided to a certain extent, so that applicable field and range are more wide.
Next, the dress that the synthesis credit score according to an embodiment of the present disclosure for being used to calculate equipment will be described referring to Fig. 7
It sets.Described device can be the server above with reference to Fig. 1 description.As shown in fig. 7, device 700 includes: acquiring unit
701, feature extraction unit 702 and comprehensive credit score computing unit 703.
Acquiring unit 701 is used to obtain the initial data about an equipment.
Feature extraction unit 702 is used to extract the feature of multiple dimensions from the initial data, wherein each dimension
Including multiple features corresponding with the dimension.
Comprehensive credit score computing unit 703 is used under each dimension, based on each feature corresponding with the dimension and
Weight corresponding with each feature determines the credit score under each dimension, and is based on weight corresponding with the multiple dimension
With the credit score under each dimension, the synthesis credit score of the equipment is determined.
As it can be seen that realizing a kind of credit rating in various dimensions to equipment by device according to an embodiment of the present disclosure
Model.Compared with being based only upon single device-fingerprint in the prior art and marking the scheme of black equipment, can more effectively it cope with
The case where equipment distorts, conflicts, protocol entry.In addition, equipment can be calculated by method according to an embodiment of the present disclosure
Synthesis credit score.With in the prior art only by 0 and 1 come the scheme of marking arrangement compared with, confidence level can be further realized
Differentiation.Also, since what is calculated in device according to an embodiment of the present disclosure is equipment rather than the comprehensive letter of user identity
With point, therefore other business sides do not need to know the details of user, do not need the customer attribute information to be synchronized to the yet
Tripartite provides for credit level based on equipment and differentiates, avoids the use of privacy of user data to a certain extent, from
And applicable field and range are more wide.
Wherein, the comprehensive credit score computing unit 703 may further include training unit (not shown).Training
Unit can be used for being trained by using the first supervised classification algorithm based on first sample library, with determine it is described each
The weight of each feature corresponding with the dimension under dimension.Also, training unit can be also used for by using the first supervision point
Class algorithm is trained based on first sample library, to determine the weight of each dimension.For example, as mentioned above it is possible, can
Each weight is determined in the method by linear regression.
As mentioned above it is possible, first sample library includes the sample of all scenes.In other words, first sample library is considered as leading to
Use sample database.It is the universal model being applicable in for all scenes based on the model that first sample library trains.However, in order to more preferable
Ground adapts to various special scenes, it is also desirable to can be various to improve on the basis of universal model further combined with special purpose model
Accuracy and discrimination under special scenes.
Therefore, as alternatively possible embodiment, as shown in figure 8, in addition to acquiring unit 701, feature extraction unit
702 with except comprehensive credit score computing unit 703, and device 800 according to another embodiment of the present disclosure can also be wrapped further
It includes: risk probability computing unit 801 and adjustment unit 802.
Risk probability computing unit 801 is used to be instructed by using the second supervised classification algorithm based on the second sample database
Practice to obtain an equipment Risk prediction model, and be based on the equipment Risk prediction model, determines the risk probability of the equipment.
As mentioned above it is possible, different from first sample library, the second sample database is the sample database for special scenes.Therefore, it is based on second
The model that sample database trains is the special purpose model for special scenes.
The risk probability for the equipment that adjustment unit 802 is used to determine based on the risk probability computing unit 801 is adjusted
The synthesis credit score for the equipment that the whole comprehensive credit score computing unit 703 determines.
Since the device according to the synthesis credit score of the calculating equipment of the disclosure can be applied to the risk control of multiple business
System processed, including but not limited to movable anti-brush, malicious registration, identity authentication, stolen protection etc..In addition, although the disclosure is with industry
Risk control be engaged in as point of penetration, but the application field of specific equipment complex credit score not only includes security risk control neck
Domain applies also for the multiple fields such as recommendation, advertisement.That is, in the biggish different application scene of otherness, by using
It is directed to the special purpose model of the application scenarios, the accuracy of final calculated comprehensive credit score can be effectively improved.
As a kind of possible embodiment, can believe only for high risk and the comprehensive of the equipment in high credible two sections
It is adjusted with point.For example, the risk probability based on the equipment, the adjustment unit can be configured to: if
The comprehensive credit score is less than first threshold and the risk probability is less than second threshold, then improves the comprehensive credit score；With
And if the comprehensive credit score is greater than third threshold value and the risk probability is greater than the 4th threshold value, reduce the comprehensive credit
Point.
Therefore, in this embodiment according to the disclosure, by using the calculated equipment based on special purpose model
Risk probability adjusts the synthesis credit score of equipment, can further increase the accuracy of the synthesis credit score of equipment, more effectively
Distinguish black equipment and white equipment in ground.
Furthermore, it is possible to it is very few in the presence of the initial data about a certain equipment, and then the characteristic under each dimension extracted
Very few situation.In this case, the synthesis credit score based on the above calculated equipment of method cannot have
Distinguish black equipment and equipment in effect ground.
It therefore, can be by establishing the network with the associated other equipment of equipment as another possible embodiment
Model, the association mining for then carrying out credit score calculate.Analysis is propagated using association by the synthesis credit score to equipment, is promoted
The accuracy rate of comprehensive credit score, so that the comprehensive credit score of equipment completion that data characteristics dimension is not complete.The association of comprehensive credit score
Propagating analysis is to be based on " being likely to be black equipment with the associated equipment of black equipment, also having very much with the associated equipment of white equipment
May be white equipment " hypothesis and carry out.
It is pointed out here that and not all equipment can be carried out the association of comprehensive credit score and propagate.Carry out
Before association is propagated, need to judge the equipment as propagating source.Specifically, Fig. 9 is shown according to the disclosure again
The functional block diagram of the device of one embodiment.As shown in figure 9, in addition to acquiring unit 701, feature extraction unit 702 and comprehensive letter
With dividing except computing unit 703, device 900 according to another embodiment of the present disclosure be can further include: association is propagated
Unit 901, for when the synthesis credit score of the equipment meets predetermined condition, the synthesis credit score based on the equipment, really
The fixed synthesis credit score with the associated other equipment of the equipment.
For example, the predetermined condition can be the synthesis credit score of the equipment in specific sections.In general, calculating
The range of the synthesis credit score of equipment out is 0-800.Wherein, if score is closer to two endpoint values (that is, 0,800), that
The discrimination of equipment is higher, if score, closer to median (that is, 400), the discrimination of equipment is lower.By discrimination
High equipment is known as firmly believing equipment, the equipment that can distinguish black equipment or white equipment.As an example, it will integrate
Credit score is regarded as less than the equipment that a threshold value (e.g., 200) and comprehensive credit score are greater than another threshold value (e.g., 600) firmly believes equipment,
And only to firmly believing that the score of equipment is associated propagation.
Certainly, the credit score of itself can not only be propagated to other associate devices by current device, but also can be received
It is propagated to the credit score from other associate devices.In this case, comprehensive credit score computing unit 703 can will calculate
Current device synthesis credit score and the credit score propagated from other associate devices it is comprehensive to obtain final synthesis credit score.
For example, can be by defeated by the synthesis credit score of calculated current device and the synthesis credit score propagated from other associate devices
Enter to a weighted model and obtains final synthesis credit score.For example, can be based on the correlation degree between two associate devices
To determine the weight of propagated credit score.
Increase the embodiment of association propagation unit on the basis of describing device shown in Fig. 7 above by reference to Fig. 9.When
So, the disclosure is not limited to that.For example, it is also possible to increase association propagation unit on the basis of device shown in Fig. 8.
Due in the device and the above method according to the synthesis credit score for calculating equipment of the disclosure
Each step is completely corresponding, thus in order to avoid redundancy for the sake of, its details is not unfolded to describe here.
An example such as figure according to the device of the synthesis credit score for calculating equipment of the disclosure as hardware entities
Shown in 10.Described device includes processor 1001, memory 1002 and at least one external communication interface 1003.The processing
Device 1001, memory 1002 and external communication interface 1003 are connected by bus 1004.
For the processor 1001 for data processing, when executing processing, microprocessor, centre can be used
Manage device (CPU, Central Processing Unit), digital signal processor (DSP, Digital Singnal
Processor) or programmable logic array (FPGA, Field-Programmable Gate Array) is realized；For storage
It include operational order, which can be computer-executable code, by the operational order come real for device 1002
Each step in the method flow of each embodiment of the existing above-mentioned disclosure.
Figure 11 shows the schematic diagram of the computer readable recording medium of embodiment according to the present invention.As shown in figure 11,
Computer readable recording medium 1100 according to an embodiment of the present invention is stored thereon with computer program instructions 1101.When the meter
When calculation machine program instruction 1101 is run by processor, the gray scale control according to an embodiment of the present invention referring to the figures above description is executed
Method processed.
So far, it is described in detail referring to figs. 1 to Figure 11 according to an embodiment of the present disclosure for calculating equipment
Synthesis credit score method and apparatus.In the method according to an embodiment of the present disclosure for calculating the synthesis credit score of equipment
In device, a kind of model in various dimensions to the credit rating of equipment is realized.Be based only upon single set in the prior art
The scheme of the standby black equipment of Finger-print labelling method is compared, can more effectively cope with equipment distort, conflict, protocol entry the case where.In addition,
By method and apparatus according to an embodiment of the present disclosure, the synthesis credit score of equipment can be calculated.With in the prior art only
It is compared by 0 with 1 come the scheme of marking arrangement, the differentiation of confidence level can be further realized.Also, due to according to the disclosure
Embodiment method in calculate is equipment rather than the synthesis credit score of user identity, therefore other business sides do not need to know
The details of road user, do not need customer attribute information to be synchronized to third party yet and provide for credit level based on equipment
Differentiate, the use of privacy of user data is avoided to a certain extent, so that applicable field and range are more wide.Separately
Outside, in the method and apparatus of synthesis credit score according to an embodiment of the present disclosure for calculating equipment, by using being based on
Special purpose model and calculated equipment Risk probability adjust the synthesis credit score of equipment, can further increase the synthesis of equipment
Black equipment and white equipment are more effectively distinguished in the accuracy of credit score.Also, by the synthesis credit score to equipment using pass
Connection propagates analysis, promotes the accuracy rate of comprehensive credit score, so that the comprehensive credit score of equipment completion that data characteristics dimension is not complete.
It should be noted that in the present specification, the terms "include", "comprise" or its any other variant are intended to
Non-exclusive inclusion, so that the process, method, article or equipment including a series of elements is not only wanted including those
Element, but also including other elements that are not explicitly listed, or further include for this process, method, article or equipment
Intrinsic element.In the absence of more restrictions, the element limited by sentence " including ... ", it is not excluded that including
There is also other identical elements in the process, method, article or equipment of the element.
Finally, it is to be noted that, it is above-mentioned it is a series of processing not only include with sequence described here in temporal sequence
The processing of execution, and the processing including executing parallel or respectively rather than in chronological order.
Through the above description of the embodiments, those skilled in the art can be understood that the disclosure can be by
Software adds the mode of required hardware platform to realize, naturally it is also possible to all be implemented by software.Based on this understanding,
The technical solution of the disclosure can be embodied in the form of software products in whole or in part to what background technique contributed,
The computer software product can store in storage medium, such as ROM/RAM, magnetic disk, CD, including some instructions are to make
It obtains a computer equipment (can be personal computer, server or the network equipment etc.) and executes each embodiment of the disclosure
Or method described in certain parts of embodiment.
The disclosure is described in detail above, used herein principle and embodiment party of the specific case to the disclosure
Formula is expounded, disclosed method that the above embodiments are only used to help understand and its core concept；Meanwhile it is right
Change is had in specific embodiments and applications according to the thought of the disclosure in those of ordinary skill in the art
Place, in conclusion the content of the present specification should not be construed as the limitation to the disclosure.
Claims (15)
1. a kind of method for calculating the synthesis credit score of equipment, comprising:
Obtain the initial data about an equipment；
The feature of multiple dimensions is extracted from the initial data, wherein each dimension includes multiple spies corresponding with the dimension
Sign；
Under each dimension, it is based on and the corresponding each feature of the dimension and weight corresponding with each feature, is determined each
Credit score under a dimension；And
Based on the credit score under weight corresponding with the multiple dimension and each dimension, the synthesis credit of the equipment is determined
Point,
Wherein the multiple dimension includes at least two in social dimension, security dimension, value relevance and primary attribute dimension.
2. according to the method described in claim 1, further comprising:
It by using the first supervised classification algorithm, is trained based on first sample library, is determined in a manner of by learning automatically
The weight of each feature corresponding with the dimension under each described dimension；And
It by using the first supervised classification algorithm, is trained based on first sample library, is determined in a manner of by learning automatically
The weight of each dimension.
3. according to the method described in claim 1, further comprising:
By using the second supervised classification algorithm, it is trained based on the second sample database to obtain an equipment Risk prediction model；
And
Based on the equipment Risk prediction model, the risk probability of the equipment is determined.
4. according to the method described in claim 3, further comprising:
Based on the risk probability of the equipment, the synthesis credit score of the equipment is adjusted.
5. according to the method described in claim 4, the risk probability wherein based on the equipment, adjusts the comprehensive credit score
Step includes:
If the comprehensive credit score is less than first threshold and the risk probability is less than second threshold, the comprehensive letter is improved
With point；And
If the comprehensive credit score is greater than third threshold value and the risk probability is greater than the 4th threshold value, the comprehensive letter is reduced
With point.
6. method according to claim 1 or 4, further comprises:
When the synthesis credit score of the equipment meets predetermined condition, the synthesis credit score based on the equipment, it is determining with it is described
The synthesis credit score of the associated other equipment of equipment.
7. according to the method described in claim 1, wherein
Under social dimension, based on multiple features of the account associated by equipment under social scene and corresponding weight come really
Determine credit score；
Under security dimension, credit is determined based on multiple features relevant to the behavior of doing evil of equipment and corresponding weight
Point；
Under value relevance, property of value feature based on equipment and corresponding weight determine credit score；And
Under primary attribute dimension, technical data feature based on equipment and corresponding weight determine credit score.
8. a kind of for calculating the device of the synthesis credit score of equipment, comprising:
Acquiring unit, for obtaining the initial data about an equipment；
Feature extraction unit, for extracting the feature of multiple dimensions from the initial data, wherein each dimension include with
The corresponding multiple features of the dimension；And
Comprehensive credit score computing unit, under each dimension, based on each feature corresponding with the dimension and with it is each
The corresponding weight of feature, determines the credit score under each dimension, and based on weight corresponding with the multiple dimension and each
Credit score under dimension determines the synthesis credit score of the equipment,
Wherein the multiple dimension includes at least two in social dimension, security dimension, value relevance and primary attribute dimension.
9. device according to claim 8, wherein the comprehensive credit score computing unit further comprises:
Training unit, for being trained based on first sample library by using the first supervised classification algorithm, by learning automatically
The mode of habit determines the weight of each feature corresponding with the dimension under each described dimension, and by using the first supervision
Sorting algorithm is trained based on first sample library, and the weight of each dimension is determined in a manner of by learning automatically.
10. device according to claim 8, further comprises:
Risk probability computing unit, for being trained based on the second sample database to obtain by using the second supervised classification algorithm
To an equipment Risk prediction model, and it is based on the equipment Risk prediction model, determines the risk probability of the equipment.
11. device according to claim 10, further comprises:
Adjustment unit, the risk probability of the equipment for being determined based on the risk probability computing unit are adjusted described comprehensive
Close the synthesis credit score for the equipment that credit score computing unit determines.
12. device according to claim 11, wherein the adjustment unit is configured to:
If the comprehensive credit score is less than first threshold and the risk probability is less than second threshold, the comprehensive letter is improved
With point；And
If the comprehensive credit score is greater than third threshold value and the risk probability is greater than the 4th threshold value, the comprehensive letter is reduced
With point.
13. the device according to claim 8 or 11, further comprises:
It is associated with propagation unit, for when the synthesis credit score of the equipment meets predetermined condition, the synthesis based on the equipment
Credit score, the determining synthesis credit score with the associated other equipment of the equipment.
14. device according to claim 8, wherein the comprehensive credit score computing unit is configured to:
Under social dimension, based on multiple features of the account associated by equipment under social scene and corresponding weight come really
Determine credit score；
Under security dimension, credit is determined based on multiple features relevant to the behavior of doing evil of equipment and corresponding weight
Point；
Under value relevance, credit score is determined based on the property of value feature of equipment itself and corresponding weight；And
Under primary attribute dimension, technical data feature based on equipment and corresponding weight determine credit score.
15. a kind of computer readable recording medium, for storing computer program on it, when executing the calculating by processor
When machine program, perform the steps of
Obtain the initial data about an equipment；
The feature of multiple dimensions is extracted from the initial data, wherein each dimension includes multiple spies corresponding with the dimension
Sign；
Under each dimension, it is based on and the corresponding each feature of the dimension and weight corresponding with each feature, is determined each
Credit score under a dimension；And
Based on the credit score under weight corresponding with the multiple dimension and each dimension, the synthesis credit of the equipment is determined
Point,
Wherein the multiple dimension includes at least two in social dimension, security dimension, value relevance and primary attribute dimension.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811212694.XA CN110162958B (en) | 2018-10-18 | 2018-10-18 | Method, apparatus and recording medium for calculating comprehensive credit score of device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811212694.XA CN110162958B (en) | 2018-10-18 | 2018-10-18 | Method, apparatus and recording medium for calculating comprehensive credit score of device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110162958A true CN110162958A (en) | 2019-08-23 |
| CN110162958B CN110162958B (en) | 2023-04-18 |
Family
ID=67645085
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811212694.XA Active CN110162958B (en) | 2018-10-18 | 2018-10-18 | Method, apparatus and recording medium for calculating comprehensive credit score of device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110162958B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110782333A (en) * | 2019-08-26 | 2020-02-11 | 腾讯科技（深圳）有限公司 | Equipment risk control method, device, equipment and medium |
| CN112488590A (en) * | 2020-12-21 | 2021-03-12 | 青岛海尔科技有限公司 | Target object classification method and device, storage medium and electronic device |
| CN113065908A (en) * | 2020-01-02 | 2021-07-02 | 中国移动通信有限公司研究院 | Rental method, rental device, rental platform, and storage medium |
| CN113469481A (en) * | 2020-03-31 | 2021-10-01 | 北京沃东天骏信息技术有限公司 | Visual processing method and equipment for evaluation data of object to be processed |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101582810A (en) * | 2008-05-15 | 2009-11-18 | 华为技术有限公司 | Secure state evaluating method, network equipment and network system |
| CN101620653A (en) * | 2008-07-04 | 2010-01-06 | 北京启明星辰信息技术股份有限公司 | System and method for evaluating security risk based on asset weak point analysis |
| CN102710598A (en) * | 2011-04-19 | 2012-10-03 | 卡巴斯基实验室封闭式股份公司 | System and method for reducing security risk in computer network |
| WO2016078388A1 (en) * | 2014-11-21 | 2016-05-26 | 中兴通讯股份有限公司 | Data aging method and apparatus |
| CN105790866A (en) * | 2016-03-31 | 2016-07-20 | 中国联合网络通信集团有限公司 | Base station grading method and base station grading device |
| CN106068513A (en) * | 2014-02-28 | 2016-11-02 | 时空防御系统有限责任公司 | Safety estimation system and method |
| CN106355405A (en) * | 2015-07-14 | 2017-01-25 | 阿里巴巴集团控股有限公司 | Method and device for identifying risks and system for preventing and controlling same |
| CN106992904A (en) * | 2017-05-19 | 2017-07-28 | 湖南省起航嘉泰网络科技有限公司 | Network equipment health degree appraisal procedure based on dynamic comprehensive weight |
| CN108416669A (en) * | 2018-03-13 | 2018-08-17 | 腾讯科技（深圳）有限公司 | User behavior data processing method, device, electronic equipment and computer-readable medium |
| CN108615101A (en) * | 2016-12-09 | 2018-10-02 | 爱信诺征信有限公司 | Credit information processing method and processing device |
-
2018
- 2018-10-18 CN CN201811212694.XA patent/CN110162958B/en active Active
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101582810A (en) * | 2008-05-15 | 2009-11-18 | 华为技术有限公司 | Secure state evaluating method, network equipment and network system |
| CN101620653A (en) * | 2008-07-04 | 2010-01-06 | 北京启明星辰信息技术股份有限公司 | System and method for evaluating security risk based on asset weak point analysis |
| CN102710598A (en) * | 2011-04-19 | 2012-10-03 | 卡巴斯基实验室封闭式股份公司 | System and method for reducing security risk in computer network |
| CN106068513A (en) * | 2014-02-28 | 2016-11-02 | 时空防御系统有限责任公司 | Safety estimation system and method |
| WO2016078388A1 (en) * | 2014-11-21 | 2016-05-26 | 中兴通讯股份有限公司 | Data aging method and apparatus |
| CN106355405A (en) * | 2015-07-14 | 2017-01-25 | 阿里巴巴集团控股有限公司 | Method and device for identifying risks and system for preventing and controlling same |
| CN105790866A (en) * | 2016-03-31 | 2016-07-20 | 中国联合网络通信集团有限公司 | Base station grading method and base station grading device |
| CN108615101A (en) * | 2016-12-09 | 2018-10-02 | 爱信诺征信有限公司 | Credit information processing method and processing device |
| CN106992904A (en) * | 2017-05-19 | 2017-07-28 | 湖南省起航嘉泰网络科技有限公司 | Network equipment health degree appraisal procedure based on dynamic comprehensive weight |
| CN108416669A (en) * | 2018-03-13 | 2018-08-17 | 腾讯科技（深圳）有限公司 | User behavior data processing method, device, electronic equipment and computer-readable medium |
Non-Patent Citations (2)
| Title |
|---|
| 王君泽: "《网络舆情应对的关键技术研究》", 31 January 2017 * |
| 高勃;祝凌曦;肖雪梅;: "城市轨道交通路网运营设备安全状态评估方法" * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110782333A (en) * | 2019-08-26 | 2020-02-11 | 腾讯科技（深圳）有限公司 | Equipment risk control method, device, equipment and medium |
| CN110782333B (en) * | 2019-08-26 | 2023-10-17 | 腾讯科技（深圳）有限公司 | Equipment risk control method, device, equipment and medium |
| CN113065908A (en) * | 2020-01-02 | 2021-07-02 | 中国移动通信有限公司研究院 | Rental method, rental device, rental platform, and storage medium |
| CN113469481A (en) * | 2020-03-31 | 2021-10-01 | 北京沃东天骏信息技术有限公司 | Visual processing method and equipment for evaluation data of object to be processed |
| CN112488590A (en) * | 2020-12-21 | 2021-03-12 | 青岛海尔科技有限公司 | Target object classification method and device, storage medium and electronic device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110162958B (en) | 2023-04-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11475143B2 (en) | Sensitive data classification | |
| WO2020253358A1 (en) | Service data risk control analysis processing method, apparatus and computer device | |
| Mohan et al. | An approach to forecast impact of Covid‐19 using supervised machine learning model | |
| CN103176983B (en) | A kind of event method for early warning based on internet information | |
| CN107025509B (en) | Decision making system and method based on business model | |
| CN110162958A (en) | For calculating the method, apparatus and recording medium of the synthesis credit score of equipment | |
| CN109241711A (en) | User behavior recognition method and device based on prediction model | |
| CN107222865A (en) | The communication swindle real-time detection method and system recognized based on suspicious actions | |
| KR102046501B1 (en) | Service providing apparatus and method for evaluating valuation and supporting upbringing based on diagnosis of start-up company | |
| CN104077396A (en) | Method and device for detecting phishing website | |
| CN113627566A (en) | Early warning method and device for phishing and computer equipment | |
| CN111460294A (en) | Message pushing method and device, computer equipment and storage medium | |
| CN107193974A (en) | Localized information based on artificial intelligence determines method and apparatus | |
| CN104915842A (en) | Electronic commerce transaction monitoring method based on internet transaction data | |
| CN103177129A (en) | Internet real-time information recommendation and prediction system | |
| WO2022142903A1 (en) | Identity recognition method and apparatus, electronic device, and related product | |
| KR101924352B1 (en) | Method for detecting issue based on trend analysis device thereof | |
| CN110162939B (en) | Man-machine identification method, equipment and medium | |
| Shao et al. | How does facial recognition as an urban safety technology affect firm performance? The moderating role of the home country’s government subsidies | |
| CN114819967A (en) | Data processing method and device, electronic equipment and computer readable storage medium | |
| CN107368499B (en) | Client label modeling and recommending method and device | |
| CN112966014A (en) | Method and device for searching target object | |
| CN113298121A (en) | Message sending method and device based on multi-data source modeling and electronic equipment | |
| CN116996325A (en) | Network security detection method and system based on cloud computing | |
| CN112990989B (en) | Value prediction model input data generation method, device, equipment and medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |