CN108985583A

CN108985583A - Finance data risk control method and device based on artificial intelligence

Info

Publication number: CN108985583A
Application number: CN201810677096.3A
Authority: CN
Inventors: 霍雨佳; 张帆; 许超
Original assignee: Bank of China Ltd
Current assignee: Bank of China Ltd
Priority date: 2018-06-27
Filing date: 2018-06-27
Publication date: 2018-12-11

Abstract

The embodiment of the invention provides a kind of finance data risk control method and device based on artificial intelligence, finance data risk control method includes: acquisition historical financial data sample, it wherein, include each Transaction Information and corresponding type of transaction label in the historical financial data sample；Based on preset rules, according to each Transaction Information and corresponding type of transaction label in the historical financial data sample, establish finance data air control model；And the type of transaction of destination financial data is predicted according to the finance data air control model, and risk control is carried out to the destination financial data according to prediction result.The present invention can effectively improve the accuracy of finance data risk control, and risk control process efficiency is high and intelligence degree is high.

Description

Finance data risk control method and device based on artificial intelligence

Technical field

The present invention relates to technical field of data processing, and in particular to a kind of finance data risk control based on artificial intelligence Method and device.

Background technique

The fast development of financial technology changes financial development mode deeply, while also having caused what financial business faced The many-sided risk to emerge one after another.Financial risks including credit intermediary, criminal, dolus malus etc. can make bank with And client is by the threat of the heavy losses of property.Therefore, it is necessary to carry out the risk control of finance data, to determine that malicious user is believed Breath, the threat of fraud that identification client encounters in service links such as payment, activity, financing, air controls reduce money loss rate, reduce enterprise Loss.

In the prior art, the mode for carrying out risk control to finance data is usually the experience accumulation according to business expert, Air control rule factor is formulated by the experience of domain expert, and risk control is carried out to finance data according to the air control rule factor System.

However, the mode scheme for artificially formulating air control rule factor not only consumes a large amount of human costs, but also it is limited to list Its rule factor of the ken of a expert's grasp is frequently not overall optimal solution, that is to say, that there are air control effects for which Rate is low, the period is long and the problem of air control result inaccuracy.

Summary of the invention

For the problems of the prior art, the present invention provides a kind of finance data risk control method based on artificial intelligence And device, the accuracy of finance data risk control can be effectively improved, risk control process efficiency is high and intelligence degree is high.

In order to solve the above technical problems, the present invention the following technical schemes are provided:

In a first aspect, the present invention provides a kind of finance data risk control method based on artificial intelligence, the finance number Include: according to risk control method

Acquire historical financial data sample, wherein include in the historical financial data sample each Transaction Information and Corresponding type of transaction label；

Based on preset rules, according in the historical financial data sample each Transaction Information and corresponding type of transaction Label establishes finance data air control model；

And the type of transaction of destination financial data is predicted according to the finance data air control model, and according to Prediction result carries out risk control to the destination financial data.

In one embodiment, it is described based on preset rules, according to each Transaction Information in the historical financial data and right The type of transaction label answered establishes finance data air control model, comprising:

The historical financial data sample is pre-processed；

Pretreated historical financial data sample is sampled, training data is obtained；

Aspect of model selection is carried out based on the training data, obtains training characteristics；

And model training is carried out to default sorting algorithm according to the training characteristics, obtain the finance data air control Model.

It is described that the historical financial data sample is pre-processed in one embodiment, comprising:

Data cleansing processing is carried out to the historical financial data sample, screens out noise data and repeated data；

And data preanalysis processing is carried out to through data cleansing treated historical financial data sample, it obtains described Data structure feature through data cleansing treated historical financial data sample.

It is described to be sampled pretreated historical financial data sample in one embodiment, training data is obtained, is wrapped It includes:

According to the data structure feature through data cleansing treated historical financial data sample, to described through data Cleaned historical financial data sample is sampled, and obtains the training data as machine learning；

It wherein, include positive example sample and negative data in the training data, and the positive example sample corresponds to first Type of transaction label, the corresponding second type of transaction label of the negative data.

It is described that aspect of model selection is carried out based on the training data in one embodiment, obtain training characteristics, comprising:

The training data is pre-processed according to the first preset rules；

And screened pretreated training data based on the second preset rules, obtain the training characteristics.

In one embodiment, first preset rules are characterized coding, feature normalization, feature discretization and Feature Dimension Reduction One of mode mode, whole modes or any combination；

Second preset rules are information gain judgment mode and/or the phase for the linear relationship between being characterized by Relationship number judgment mode.

It is described that model training is carried out to default sorting algorithm according to the training characteristics in one embodiment, obtain the gold Melt data air control model, comprising:

The input for constructing the default sorting algorithm according to each training characteristics and corresponding type of transaction label is special Levy sequence；

And model training is carried out to described pair of default sorting algorithm based on input feature vector sequence, obtain the financial number According to air control model；

Wherein, the default sorting algorithm is decision Tree algorithms or random forests algorithm.

It is described to be carried out in advance according to type of transaction of the finance data air control model to destination financial data in one embodiment It surveys, and risk control is carried out to the destination financial data according to prediction result, comprising:

The type of transaction of destination financial data is predicted according to the finance data air control model, prediction obtains described The corresponding type of transaction label of destination financial data is the first type of transaction label or the second type of transaction label；

And if the type of transaction label is the first type of transaction label, export corresponding first default risk control Result processed；

If the type of transaction label is the second type of transaction label, corresponding second default risk control knot is exported Fruit.

In one embodiment, before the acquisition historical financial data sample, the finance data risk control method is also Include:

After the completion of each process of exchange, the corresponding Transaction Information of each process of exchange is obtained；

And the corresponding type of transaction mark of each Transaction Information is generated according to the corresponding type of transaction of each process of exchange Label；

Each Transaction Information is stored into historical financial data sample database, wherein the historical financial number According to the corresponding relationship being stored in sample database between each Transaction Information and type of transaction label.

In one embodiment, the historical financial data sample database is distributed data base, and the distributed data base It is stored in block chain network.

Second aspect, the present invention provide a kind of finance data risk control system based on artificial intelligence, the finance number Include: according to risk control system

Historical financial data sample collection module, for acquiring historical financial data sample, wherein the historical financial number According to including each Transaction Information and corresponding type of transaction label in sample；

Finance data air control model construction module, for being based on preset rules, according in the historical financial data sample Each Transaction Information and corresponding type of transaction label, establish finance data air control model；

Risk control module, for being carried out according to type of transaction of the finance data air control model to destination financial data Prediction, and risk control is carried out to the destination financial data according to prediction result.

The third aspect, the present invention provides a kind of electronic equipment, including memory, processor and storage are on a memory and can The computer program run on a processor, the processor realize the gold based on artificial intelligence when executing described program The step of melting data risk control method.

Fourth aspect, the present invention provide a kind of computer readable storage medium, are stored thereon with computer program, the calculating The step of finance data risk control method based on artificial intelligence is realized when machine program is executed by processor.

As shown from the above technical solution, it is real to provide a kind of finance data risk control method based on artificial intelligence by the present invention Apply example, pass through acquisition historical financial data sample, wherein include in the historical financial data sample each Transaction Information and Corresponding type of transaction label；Based on preset rules, according to each Transaction Information in the historical financial data sample and right The type of transaction label answered establishes finance data air control model；And according to the finance data air control model to destination financial The type of transaction of data is predicted, and carries out risk control to the destination financial data according to prediction result, by directly adopting Collection establishes finance data air control model with the historical financial data sample of type of transaction label, can effectively improve finance data The building efficiency of air control model is predicted, energy according to intelligent data of the finance data air control model realization to destination financial data The forecasting efficiency and accuracy rate of prediction result are enough effectively improved, and then the accurate of finance data risk control can be effectively improved Property, risk control process efficiency is high and intelligence degree is high, also can satisfy the target for needing to carry out finance data risk control The user demand of user simultaneously improves user experience, and target user is enabled to accurately identify malicious user information, and identification client exists The threat of fraud that the service links such as payment, activity, financing, air control encounter reduces money loss rate, reduces the loss of enterprise.

Detailed description of the invention

In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, to embodiment or will show below There is attached drawing needed in technical description to be briefly described, it should be apparent that, the accompanying drawings in the following description is the present invention Some embodiments for those of ordinary skill in the art without creative efforts, can also basis These attached drawings obtain other attached drawings.

Fig. 1 is the flow diagram of the finance data risk control method embodiment of the invention based on artificial intelligence.

Fig. 2 is the information exchange schematic diagram of finance data risk control method embodiment of the invention.

Fig. 3 is the idiographic flow schematic diagram of step 200 in finance data risk control method embodiment of the invention.

Fig. 4 is the data processing schematic diagram in the step 200 in finance data risk control method embodiment of the invention.

Fig. 5 is the idiographic flow schematic diagram of step 201 in finance data risk control method embodiment of the invention.

Fig. 6 is the idiographic flow schematic diagram of step 203 in finance data risk control method embodiment of the invention.

Fig. 7 is the idiographic flow schematic diagram of step 204 in finance data risk control method embodiment of the invention.

Fig. 8 is the idiographic flow schematic diagram of step 300 in finance data risk control method embodiment of the invention.

Fig. 9 is the data processing schematic diagram in the step 300 in finance data risk control method embodiment of the invention.

Figure 10 is the detailed process signal of step 001 to 003 in finance data risk control method embodiment of the invention Figure.

Figure 11 is the data processing in the step 001 to 003 in finance data risk control method embodiment of the invention Schematic diagram.

Figure 12 is the information exchange schematic diagram of finance data risk control method of the invention.

Figure 13 is the flow diagram of the finance data risk control method application example of the invention based on artificial intelligence.

Figure 14 is model training and model verifying signal in finance data risk control method application example of the invention Figure.

Figure 15 is the intelligent predicting schematic diagram in finance data risk control method application example of the invention.

Figure 16 is the structural schematic diagram of the finance data risk control system embodiment of the invention based on artificial intelligence.

Figure 17 is that another structure of the finance data risk control system embodiment of the invention based on artificial intelligence is shown It is intended to.

Figure 18 is the kind structural schematic diagram of electronic equipment embodiment of the invention.

Specific embodiment

In order to make the object, technical scheme and advantages of the embodiment of the invention clearer, below in conjunction with the embodiment of the present invention In attached drawing, the technical scheme in the embodiment of the invention is clearly and completely described, it is clear that described embodiment is A part of the embodiment of the present invention, instead of all the embodiments.Based on the embodiments of the present invention, those of ordinary skill in the art Every other embodiment obtained without creative efforts, shall fall within the protection scope of the present invention.

The embodiment of the present invention provides a kind of specific embodiment party of finance data risk control method based on artificial intelligence Formula, referring to Fig. 1 and Fig. 2, the finance data risk control method based on artificial intelligence specifically include following content:

Step 100: acquisition historical financial data sample, wherein include each friendship in the historical financial data sample Easy information and corresponding type of transaction label.

In step 100, the finance data risk control system based on artificial intelligence acquires historical financial data sample This, for constructing finance data air control model.It is understood that the finance data risk control system based on artificial intelligence System can be presented as a kind of server, can also in the hardware composition of the finance data risk control system based on artificial intelligence To include terminal device, the terminal device can have display function.Specifically, the terminal device may include intelligent hand Machine, Flat electronic equipment, network machine top box, portable computer, desktop computer, personal digital assistant (PDA), mobile unit, Intelligent wearable device etc..Wherein, the intelligent wearable device may include smart glasses, smart watches, Intelligent bracelet etc..

The server can be communicated with the terminal device.It can be between the server and the terminal device It is communicated using any suitable network protocol, including the network protocol not yet developed in the application submitting day.The net Network agreement for example may include ICP/IP protocol, UDP/IP agreement, http protocol, HTTPS agreement etc..Certainly, the network association View for example can also include used on above-mentioned agreement RPC agreement (Remote Procedure Call Protocol, far Journey invocation of procedure agreement), REST agreement (Representational State Transfer, declarative state transfer protocol) Deng.

In the foregoing description, it for destination financial system, needs to be met according to user demand and specific application environment foundation Therefore the finance data air control model of destination financial system for each destination financial system, is required to acquire target gold Melt the historical financial data sample of system, and is exclusively used in this kind for the foundation of the historical financial data sample of the destination financial system The finance data air control model of destination financial system, to further increase the needle of the intelligent data prediction to destination financial data To property and accuracy.In a kind of concrete example, the finance data risk control system acquires certain financial institution's air control system The historical trading data sample of function test environment.

It is understood that the type of transaction tagging user shows the type of transaction of historical financial data sample, and root According to the specific requirements of destination financial system, the type of transaction label can be set to two classes and two classes or more, and subsequent gold Melting the prediction result that data air control model exports is a certain kind in all kinds of type of transaction labels, that is to say, that described Practical finance data air control model is a kind of for predicting the disaggregated model of the corresponding type of transaction label of finance data, Ke Yigen According to the data of input, type of transaction label corresponding to the data is exported.In a kind of citing, the friendship of the destination financial system There are two types of easy type labels, respectively " arm's length dealing " label and " fraudulent trading " label, to further provide for data prediction Accuracy, the type of transaction label of the destination financial system in addition to aforementioned two kinds, can also include " leave a question open transaction " label, " non-fraud but Fail Transaction " label etc., to further increase the intelligence degree of finance data risk control.

Step 200: based on preset rules, according to each Transaction Information in the historical financial data sample and corresponding Type of transaction label establishes finance data air control model.

In step 200, after collecting the historical financial data sample for constructing finance data air control model, institute Finance data risk control system is stated to continue to believe based on preset rules, according to each transaction in the historical financial data sample Breath and corresponding type of transaction label, establish finance data air control model.It is understood that the Transaction Information is for indicating In the status information of the process of exchange of destination financial system, the Transaction Information can wrap to be believed financial user containing financial user Breath, financial user the status informations such as account information, type of transaction, time, the amount of money.

Step 300: the type of transaction of destination financial data being predicted according to the finance data air control model, and root It is predicted that result carries out risk control to the destination financial data.

In step 300, after constructing the finance data air control model, the finance data risk control system pair Outer opening, the authorized person for then receiving destination financial system is sending air control instruction, and is instructed according to the air control and start basis The finance data air control model predicts the type of transaction of destination financial data, obtains the transaction class of destination financial data Type label, and then risk control is carried out to the destination financial data according to the type of transaction label.

As can be seen from the above description, the finance data risk control side based on artificial intelligence that the embodiment of the present invention provides Method establishes finance data air control model, Neng Gouyou by directly acquiring the historical financial data sample with type of transaction label Effect improves the building efficiency of finance data air control model, according to finance data air control model realization to the intelligence of destination financial data Change number it was predicted that the forecasting efficiency and accuracy rate of prediction result can be effectively improved, and then finance data wind can be effectively improved The accuracy nearly controlled, risk control process efficiency is high and intelligence degree is high, also can satisfy and needs to carry out finance data wind The user demand of the target user nearly controlled simultaneously improves user experience, and target user is enabled to accurately identify malicious user letter Breath, the threat of fraud that identification client encounters in service links such as payment, activity, financing, air controls reduce money loss rate, reduce enterprise Loss.

In a specific embodiment, the present invention also provides steps 200 in the finance data risk control method Specific embodiment, referring to Fig. 3 and Fig. 4, the step 200 specifically includes following content:

Step 201: the historical financial data sample is pre-processed.

In step 201, referring to Fig. 5, the step 201 also specifically includes following content:

Step 201a: data cleansing processing is carried out to the historical financial data sample, screens out noise data and repeat number According to.

Step 201b: data preanalysis processing is carried out to through data cleansing treated historical financial data sample, is obtained The data structure feature through data cleansing treated historical financial data sample.

It is understood that the finance data risk control system removes incredible sample by data cleansing；It goes Except meaningless data item etc.；And data general condition and data distribution are understood by data preanalysis.

Step 202: pretreated historical financial data sample being sampled, training data is obtained.

In step 202, the step 202 also specifically includes following content:

The finance data risk control system is according to described through data cleansing treated historical financial data sample Data structure feature, through data cleansing, treated that historical financial data sample is sampled to described, obtains as engineering The training data of habit；It wherein, include positive example sample and negative data in the training data, and the positive example sample is corresponding First type of transaction label, the corresponding second type of transaction label of the negative data.It is understood that the finance data wind Dangerous control system selects the training data of fraud sample and non-fraud sample as machine learning respectively.

In the foregoing description, the first type of transaction label is " arm's length dealing " label, the second type of transaction mark Label are " fraudulent trading " label.It is understood that the type of transaction label can also include third type of transaction label And the 4th type of transaction label etc., and the third type of transaction label can be " leave a question open transaction " label, and, the 4th trades Type label can be " non-fraud but Fail Transaction " label etc..

Step 203: aspect of model selection being carried out based on the training data, obtains training characteristics.

In step 203, referring to Fig. 6, the step 203 also specifically includes following content:

Step 203a: the training data is pre-processed according to the first preset rules.

Wherein, first preset rules be characterized coding, feature normalization, feature discretization and Feature Dimension Reduction mode One of mode, whole mode or any combination.It is understood that it is described according to the first preset rules to the trained number Be characterized the treatment process of engineering according to pretreatment is carried out, the Feature Engineering specifically includes: feature coding, feature normalization, Feature discretization and Feature Dimension Reduction etc..

Step 203b: being screened pretreated training data based on the second preset rules, and it is special to obtain the training Sign.

Wherein, second preset rules are for information gain judgment mode and/or for the linear pass between being characterized by The related coefficient judgment mode of system.

It is understood that if second preset rules are information gain judgment mode, the finance data risk Control system calculates the preliminary classification entropy of all features, calculates and carries out sorted entropy by a certain feature, twice the difference of entropy It is exactly the information gain of this feature.The information gain of feature is bigger, and significance level is higher；The big feature of information gain is protected It stays, abandon the small feature of information gain, be used for model training.

If second preset rules are related coefficient judgment mode, if the related coefficient of certain features is bigger, explanation There are stronger linear relationships for feature.The information that feature with strong linear relationship contains is redundancy for model training, house The efficiency of model training can be improved in the feature for abandoning strong linear relationship.

In addition, the finance data risk control system can be judged respectively with information gain judgment mode and related coefficient Mode is screened twice to by pretreated training data, obtains the training characteristics.To further increase Feature Selection Accuracy.

Step 204: model training being carried out to default sorting algorithm according to the training characteristics, obtains the finance data wind Control model.

In step 204, referring to Fig. 7, the step 204 also specifically includes following content:

Step 204a: the default sorting algorithm is constructed according to each training characteristics and corresponding type of transaction label Input feature vector sequence.

Step 204b: model training is carried out to described pair of default sorting algorithm based on input feature vector sequence, obtains the gold Melt data air control model.Wherein, the default sorting algorithm is decision Tree algorithms or random forests algorithm.

It is understood that the finance data risk control system uses decision tree, random forest scheduling algorithm training intelligence Energy air control model cheats class transaction for identification.Feature specially is selected using the feature through the feature selecting in step 203 as whole, And extract the type of transaction label (arm's length dealing, fraudulent trading) for selecting feature to correspond to original transaction data eventually；Feature is selected according to whole With the list entries of type of transaction label building decision tree, random forest scheduling algorithm；Using input feature vector sequence using decision tree, The classification based training that random forests algorithm is made whether as fraudulent trading finally obtains and judges whether a transaction is to take advantage of for the fact Cheat the intelligent air control model of transaction.

As can be seen from the above description, the finance data risk control side based on artificial intelligence that the embodiment of the present invention provides Method establishes finance data air control model, Neng Gouyou by directly acquiring the historical financial data sample with type of transaction label Effect improves building efficiency, reliability and the accuracy of finance data air control model.

In a specific embodiment, the present invention also provides steps 300 in the finance data risk control method Specific embodiment, referring to Fig. 8 and Fig. 9, the step 300 specifically includes following content:

Step 301: the type of transaction of destination financial data being predicted according to the finance data air control model, is predicted Obtaining the corresponding type of transaction label of the destination financial data is the first type of transaction label or the second type of transaction label.

Step 302: if the type of transaction label is the first type of transaction label, exporting corresponding first default risk Control result.

Step 303: if the type of transaction label is the second type of transaction label, exporting corresponding second default risk Control result.

In the foregoing description, the first type of transaction label is " arm's length dealing " label, corresponding first default risk Control result is then " refusal " movement, and the second type of transaction label is " fraudulent trading " label, corresponding second default wind Dangerous control result is then " clearance " movement.It is understood that the type of transaction label can also include third transaction class Type label and the 4th type of transaction label etc., and the third type of transaction label is " leave a question open transaction " label and the 4th transaction class Whens type label is " non-fraud but Fail Transaction " label etc., can export for third preset risk control as a result, and third Default risk control result can be movement " to be determined ", sentence so that destination financial system is secondary to destination financial data progress It is fixed.In a kind of citing, it is clear to carry out data for each sample that the finance data risk control system sends online request It washes, feature selecting, Missing Data Filling, the processes such as feature coding predict input feature vector by training pattern, to " normal to hand over It easily " makes " clearance " or makes the decision of " refusal " to " fraudulent trading ", intelligent air control system just has artificial intelligence at this time Judgement.

As can be seen from the above description, the finance data risk control side based on artificial intelligence that the embodiment of the present invention provides Method, can substitution service expert carry out customer action whether be fraud judgement, can largely reduce human cost, capture fraud Behavior improves predictablity rate, reduces the loss of assets of enterprise and client.

In a specific embodiment, the present invention also provides in the finance data risk control method in step 100 The step 001 executed before to step 003 specific embodiment, referring to Figure 10 and Figure 11, the step 001 to step 003 Specifically include following content:

Step 001: after the completion of each process of exchange, obtaining the corresponding Transaction Information of each process of exchange.

Step 002: the corresponding type of transaction of each Transaction Information is generated according to the corresponding type of transaction of each process of exchange Label.

In step 002, corresponding type of transaction label is generated in the storage of historical financial data sample, effectively to mention The high subsequent efficiency to data processing and identification.

Step 003: each Transaction Information being stored into historical financial data sample database, wherein described to go through The corresponding relationship being stored in history finance data sample database between each Transaction Information and type of transaction label.

In step 003, the historical financial data sample database is distributed data base, and the distributed data base It is stored in block chain network.

As can be seen from the above description, the finance data risk control side based on artificial intelligence that the embodiment of the present invention provides Method can effectively ensure that building gold by providing storage mode that is a kind of reliable and can not distorting for historical financial data sample Melt the accuracy of the data basis of data air control model, and then the accuracy and control of finance data risk control can be effectively improved Efficiency processed.

For further instruction this programme, the present invention also provides a kind of finance data risk control side based on artificial intelligence The specific application example of method, referring to Figure 12 and Figure 13, the finance data risk control method based on artificial intelligence it is specific Application example specifically includes following content:

The sample comprising fraud with non-fraud is obtained first from distributed data base.It is selected at random from positive negative sample respectively Take part sample as training, remaining verifies model up to standard, carry out online prediction as verifying.It is described to be based on artificial intelligence Finance data risk control method comprising the following steps:

S1: training sample of the sample of acquisition air control System Functional Test environment tape label as intelligent air control system, from The data of database derived type structure are stored.

S2: data cleansing is carried out to training sample data, main includes removing insincere sample and meaningless or repetition Data item etc..

S3: Feature Engineering is carried out to the sample after data cleansing, mainly including Feature Selection, Missing Data Filling etc., then Feature coding is carried out, referring to table 1, by feature vector qualitatively and quantitatively, training sample models primitive character item, by spy The vector of fixed dimension is converted to after assemble-publish code, and the m* being made of training data is just obtained for entire training set in this way N matrix (being laterally n feature, longitudinal m training sample)；The label data in sample is extracted, the column vector of m dimension is obtained.

Table 1

S4: the matrix and column vector that preceding step is obtained carry out machine learning training as parameter input sorting algorithm, Disaggregated model can be obtained after training, used for prediction steps.

S5: data cleansing, feature extraction etc. are carried out to the new data of online request, obtain feature vector, inputs training mould Type obtains prediction result.

Wherein, training sample of the sample of air control System Functional Test environment tape label as intelligent air control system is acquired, Including the data item of the description customer action such as terminal iidentification, certificate information, data storage, the original as machine learning are carried out Beginning sample.

Sample acquisition data item has carried out data preanalysis, data cleansing, missing to training sample as sample characteristics Value filling.From miss rate, business is angularly set out, the characteristic item of screening description customer action.

Referring to Figure 14, the method for model training in application example of the present invention is described in detail: above-mentioned steps are obtained Matrix and column vector as parameter input sorting algorithm carry out machine learning training data, training when randomly selected from data set Partial data is as training set, and remaining data is as verifying collection.Workable sorting algorithm includes: decision tree, random forest, instruction Prediction model can be obtained after white silk.

Then obtained model is verified using verifying collection, that is, checks the accuracy rate performance for having obtained model, selected The model for meeting index (accuracy rate is higher) is used for subsequent prediction link.

In addition, the use of intelligent predicting link provides prediction service by hypertext transfer protocol mode, it is distributed using open source Formula service framework accesses intelligent predicting module by hypertext transfer protocol mode, referring to Figure 15, in application example of the present invention The method of intelligent predicting is described in detail:

For the online request that open source distributed service framework is sent, the json message for standard is requested, it is right first Test sample carries out data cleansing, and the processes such as feature extraction and feature coding, the data for obtaining vectorization indicate, its is defeated Enter in the multiple models obtained to previous step training, the decision of " clearance " or " refusal " that available each model is made is led to It crosses Delphi experts investigation method method and determines optimum prediction as a result, result is transmitted to requesting party in the form of json message.

As can be seen from the above description, the finance data risk control side based on artificial intelligence that application example of the invention provides Method provides low cost, high efficiency, the air control method based on artificial intelligence of high accuracy.By to existing historic customer row For the study and training of data, program is allowed to have the judgement similar with human brain.Traditional air control system relies on business expert Rule needs to cooperate a large amount of expertise, and consumes a large amount of human costs；It can be learnt automatically according to historical data, generate intelligence Energy air control decision tree captures fraud customer action, is complementary to one another, can be effectively dropped with Expert Rules system to a certain extent Low human cost can accurately identify malicious user information to a certain degree, identify client in industry such as payment, activity, financing, air controls The threat of fraud that business link encounters reduces money loss rate, reduces the loss of enterprise；Very low to applicable environmental requirement, migration is good, Windows Unix/Linux environment can be generally applicable in.

The embodiment of the present invention, which provides, a kind of can be realized the finance data risk control method based on artificial intelligence The finance data risk control system based on artificial intelligence specific embodiment, referring to Figure 16, it is described be based on artificial intelligence Finance data risk control system specifically include following content:

Historical financial data sample collection module 10, for acquiring historical financial data sample, wherein the historical financial It include each Transaction Information and corresponding type of transaction label in data sample.

Finance data air control model construction module 20, for being based on preset rules, according to the historical financial data sample In each Transaction Information and corresponding type of transaction label, establish finance data air control model.

Risk control module 30, for according to the finance data air control model to the type of transaction of destination financial data into Row prediction, and risk control is carried out to the destination financial data according to prediction result.

The embodiment of finance data risk control system provided by the present application based on artificial intelligence specifically can be used for holding The process flow of the embodiment of the above-mentioned finance data risk control method based on artificial intelligence of row, function are no longer superfluous herein It states, is referred to the detailed description of above method embodiment.

As can be seen from the above description, the finance data risk control system based on artificial intelligence that the embodiment of the present invention provides System, establishes finance data air control model, Neng Gouyou by directly acquiring the historical financial data sample with type of transaction label Effect improves the building efficiency of finance data air control model, according to finance data air control model realization to the intelligence of destination financial data Change number it was predicted that the forecasting efficiency and accuracy rate of prediction result can be effectively improved, and then finance data wind can be effectively improved The accuracy nearly controlled, risk control process efficiency is high and intelligence degree is high, also can satisfy and needs to carry out finance data wind The user demand of the target user nearly controlled simultaneously improves user experience, and target user is enabled to accurately identify malicious user letter Breath, the threat of fraud that identification client encounters in service links such as payment, activity, financing, air controls reduce money loss rate, reduce enterprise Loss.

In a specific embodiment, the present invention also provides finance data wind in the finance data risk control system The specific embodiment of model construction module 20 is controlled, the finance data air control model construction module 20 specifically includes in following Hold:

Historical financial data sample pretreatment unit 21, for being pre-processed to the historical financial data sample.

In historical financial data sample pretreatment unit 21, the also specific packet of the historical financial data sample pretreatment unit 21 Include following content:

Data cleansing subelement 21a screens out noise for carrying out data cleansing processing to the historical financial data sample Data and repeated data.

Data preanalysis subelement 21b, for carrying out data to through data cleansing treated historical financial data sample Preanalysis processing obtains the data structure feature through data cleansing treated historical financial data sample.

Training data acquiring unit 22 is trained for pretreated historical financial data sample to be sampled Data.

In training data acquiring unit 22, the training data acquiring unit 22 also specifically includes following content:

The training data acquiring unit 22 is according to the number through data cleansing treated historical financial data sample According to structure feature, through data cleansing, treated that historical financial data sample is sampled to described, obtains as machine learning Training data；It wherein, include positive example sample and negative data in the training data, and the positive example sample corresponding the One type of transaction label, the corresponding second type of transaction label of the negative data.It is understood that the finance data risk Control system selects the training data of fraud sample and non-fraud sample as machine learning respectively.

Aspect of model selecting unit 23 obtains training characteristics for carrying out aspect of model selection based on the training data.

In aspect of model selecting unit 23, the aspect of model selecting unit 23 also specifically includes following content:

Training data pre-processes subelement 23a, for being pre-processed according to the first preset rules to the training data.

Training data screens subelement 23b, for being sieved pretreated training data based on the second preset rules Choosing, obtains the training characteristics.

Finance data air control model construction unit 24, for carrying out model to default sorting algorithm according to the training characteristics Training, obtains the finance data air control model.

In finance data air control model construction unit 24, the also specific packet of the finance data air control model construction unit 24 Include following content:

Input feature vector sequence construct subelement 24a, for according to each training characteristics and corresponding type of transaction mark Label construct the input feature vector sequence of the default sorting algorithm.

Model training subelement 24b, for carrying out model instruction to described pair of default sorting algorithm based on input feature vector sequence Practice, obtains the finance data air control model.Wherein, the default sorting algorithm is decision Tree algorithms or random forests algorithm.

It is understood that the model training subelement 24b uses decision tree, random forest scheduling algorithm training smart wind Control model cheats class transaction for identification.Feature specially is selected using the feature through the feature selecting in step 203 as whole, and is mentioned Take the type of transaction label (arm's length dealing, fraudulent trading) for selecting feature to correspond to original transaction data eventually；Feature and friendship are selected according to whole Easy type label constructs the list entries of decision tree, random forest scheduling algorithm；Decision tree, random is utilized using input feature vector sequence The classification based training that forest algorithm is made whether as fraudulent trading finally obtains and judges whether a transaction is that fraud is handed over for the fact Easy intelligent air control model.

As can be seen from the above description, the finance data risk control system based on artificial intelligence that the embodiment of the present invention provides System, establishes finance data air control model, Neng Gouyou by directly acquiring the historical financial data sample with type of transaction label Effect improves building efficiency, reliability and the accuracy of finance data air control model.

In a specific embodiment, the present invention also provides the finance data risk control system risks to control mould The specific embodiment of block 30, the risk control module 30 specifically include following content:

Predicting unit 31, it is pre- for being carried out according to type of transaction of the finance data air control model to destination financial data It surveys, it is the first type of transaction label or the second type of transaction that prediction, which obtains the corresponding type of transaction label of the destination financial data, Label.

First default risk control result output unit 32, if being the first type of transaction mark for the type of transaction label Label, then export corresponding first default risk control result.

Second default risk control result output unit 33, if being the second type of transaction mark for the type of transaction label Label, then export corresponding second default risk control result.

As can be seen from the above description, the finance data risk control system based on artificial intelligence that the embodiment of the present invention provides System, can substitution service expert carry out customer action whether be fraud judgement, can largely reduce human cost, capture fraud Behavior improves predictablity rate, reduces the loss of assets of enterprise and client.

It in a specific embodiment, include transaction letter the present invention also provides the finance data risk control system Breath obtains module 01, the specific embodiment of type of transaction tag generation module 02 and Transaction Information memory module 03, referring to figure 17, the Transaction Information obtains module 01, type of transaction tag generation module 02 and Transaction Information memory module 03 and specifically includes Following content:

Transaction Information obtains module 01, corresponding for after the completion of each process of exchange, obtaining each process of exchange Transaction Information.

Type of transaction tag generation module 02, for generating each transaction according to the corresponding type of transaction of each process of exchange The corresponding type of transaction label of information.

In type of transaction tag generation module 02, corresponding transaction class is generated in the storage of historical financial data sample Type label, to effectively improve the subsequent efficiency to data processing and identification.

Transaction Information memory module 03, for storing each Transaction Information to historical financial data sample database In, wherein it is stored in the historical financial data sample database between each Transaction Information and type of transaction label Corresponding relationship.

In Transaction Information memory module 03, the historical financial data sample database is distributed data base, and should Distributed data base is stored in block chain network.

As can be seen from the above description, the finance data risk control system based on artificial intelligence that the embodiment of the present invention provides System can effectively ensure that building gold by providing storage mode that is a kind of reliable and can not distorting for historical financial data sample Melt the accuracy of the data basis of data air control model, and then the accuracy and control of finance data risk control can be effectively improved Efficiency processed.

The embodiment of the present invention provides the finance data risk based on artificial intelligence that can be realized in above-described embodiment one The specific embodiment of a kind of electronic equipment of Overall Steps in control method, referring to Figure 18, the electronic equipment is specifically included Following content:

Processor (processor) 601, memory (memory) 602, communication interface (Communications Interface) 603 and bus 604；

Wherein, the processor 601, memory 602, communication interface 603 are completed mutual by the bus 1204 Communication；The communication interface 603 is for realizing between the relevant devices such as user terminal, destination financial system and block chain network Information transmission；

The processor 601 is used to call the computer program in the memory 602, and the processor executes the meter The Overall Steps in above-described embodiment one are realized when calculation machine program, for example, reality when the processor executes the computer program Existing following step:

As can be seen from the above description, the electronic equipment that the embodiment of the present invention provides, by directly acquiring with type of transaction The historical financial data sample of label establishes finance data air control model, can effectively improve the building of finance data air control model Efficiency is predicted according to intelligent data of the finance data air control model realization to destination financial data, can effectively improve prediction As a result forecasting efficiency and accuracy rate, and then the accuracy of finance data risk control can be effectively improved, risk control process High-efficient and intelligence degree is high, also can satisfy and needs to carry out the user demand of the target user of finance data risk control simultaneously User experience is improved, target user is enabled to accurately identify malicious user information, identifies client in payment, activity, financing, wind The threat of fraud that the service links such as control encounter reduces money loss rate, reduces the loss of enterprise.

The embodiment of the present invention provides the finance data risk based on artificial intelligence that can be realized in above-described embodiment one A kind of computer readable storage medium of Overall Steps in control method is stored with calculating on the computer readable storage medium Machine program, the computer program realize the Overall Steps of above-described embodiment one when being executed by processor, for example, the processor is held Following step is realized when the row computer program:

As can be seen from the above description, the computer readable storage medium that the embodiment of the present invention provides, by directly acquiring band There is the historical financial data sample of type of transaction label to establish finance data air control model, finance data air control can be effectively improved The building efficiency of model is predicted, Neng Gouyou according to intelligent data of the finance data air control model realization to destination financial data Effect improves the forecasting efficiency and accuracy rate of prediction result, and then can effectively improve the accuracy of finance data risk control, wind Dangerous control process is high-efficient and intelligence degree is high, also can satisfy the target user's for needing to carry out finance data risk control User demand simultaneously improves user experience, and target user is enabled to accurately identify malicious user information, and identification client is paying, living The threat of fraud that the service links such as dynamic, financing, air control encounter reduces money loss rate, reduces the loss of enterprise.

All the embodiments in this specification are described in a progressive manner, same and similar portion between each embodiment Dividing may refer to each other, and each embodiment focuses on the differences from other embodiments.Especially for hardware+ For program class embodiment, since it is substantially similar to the method embodiment, so being described relatively simple, related place is referring to side The part of method embodiment illustrates.

It is above-mentioned that this specification specific embodiment is described.Other embodiments are in the scope of the appended claims It is interior.In some cases, the movement recorded in detail in the claims or step can be come according to the sequence being different from embodiment It executes and desired result still may be implemented.In addition, process depicted in the drawing not necessarily require show it is specific suitable Sequence or consecutive order are just able to achieve desired result.In some embodiments, multitasking and parallel processing be also can With or may be advantageous.

Although this application provides the method operating procedure as described in embodiment or flow chart, based on conventional or noninvasive The labour for the property made may include more or less operating procedure.The step of enumerating in embodiment sequence is only numerous steps One of execution sequence mode, does not represent and unique executes sequence.It, can when device or client production in practice executes To execute or parallel execute (such as at parallel processor or multithreading according to embodiment or method shown in the drawings sequence The environment of reason).

Controller can be implemented in any suitable manner, for example, controller can take such as microprocessor or processing The computer for the computer readable program code (such as software or firmware) that device and storage can be executed by (micro-) processor can Read medium, logic gate, switch, specific integrated circuit (Application Specific Integrated Circuit, ASIC), the form of programmable logic controller (PLC) and insertion microcontroller, the example of controller includes but is not limited to following microcontroller Device: ARC 625D, Atmel AT91SAM, Microchip PIC18F26K20 and Silicone Labs C8051F320 are deposited Memory controller is also implemented as a part of the control logic of memory.It is also known in the art that in addition to Pure computer readable program code mode is realized other than controller, can be made completely by the way that method and step is carried out programming in logic Controller is obtained to come in fact in the form of logic gate, switch, specific integrated circuit, programmable logic controller (PLC) and insertion microcontroller etc. Existing identical function.Therefore this controller is considered a kind of hardware component, and to including for realizing various in it The device of function can also be considered as the structure in hardware component.Or even, it can will be regarded for realizing the device of various functions For either the software module of implementation method can be the structure in hardware component again.

System, device, module or the unit that above-described embodiment illustrates can specifically realize by computer chip or entity, Or it is realized by the product with certain function.It is a kind of typically to realize that equipment is computer.Specifically, computer for example may be used Think personal computer, laptop computer, vehicle-mounted human-computer interaction device, cellular phone, camera phone, smart phone, individual Digital assistants, media player, navigation equipment, electronic mail equipment, game console, tablet computer, wearable device or The combination of any equipment in these equipment of person.

Although this specification embodiment provides the method operating procedure as described in embodiment or flow chart, based on conventional It may include either more or less operating procedure without creative means.The step of being enumerated in embodiment sequence be only One of numerous step execution sequence mode does not represent and unique executes sequence.Device or end product in practice is held When row, can be executed according to embodiment or method shown in the drawings sequence or it is parallel execute (such as parallel processor or The environment of multiple threads, even distributed data processing environment).The terms "include", "comprise" or its any other change Body is intended to non-exclusive inclusion, so that process, method, product or equipment including a series of elements are not only wrapped Those elements are included, but also including other elements that are not explicitly listed, or further includes for this process, method, product Or the element that equipment is intrinsic.In the absence of more restrictions, being not precluded is including process, the side of the element There is also other identical or equivalent elements in method, product or equipment.

For convenience of description, it is divided into various modules when description apparatus above with function to describe respectively.Certainly, implementing this The function of each module can be realized in the same or multiple software and or hardware when specification embodiment, it can also be by reality Show the module of same function by the combination realization etc. of multiple submodule or subelement.Installation practice described above is only Schematically, for example, the division of the unit, only a kind of logical function partition, can there is other draw in actual implementation The mode of dividing, such as multiple units or components can be combined or can be integrated into another system, or some features can be ignored, Or it does not execute.Another point, shown or discussed mutual coupling, direct-coupling or communication connection can be by one The indirect coupling or communication connection of a little interfaces, device or unit can be electrical property, mechanical or other forms.

It is also known in the art that other than realizing controller in a manner of pure computer readable program code, it is complete Entirely can by by method and step carry out programming in logic come so that controller with logic gate, switch, specific integrated circuit, programmable Logic controller realizes identical function with the form for being embedded in microcontroller etc..Therefore this controller is considered one kind Hardware component, and the structure that the device for realizing various functions that its inside includes can also be considered as in hardware component.Or Person even, can will be considered as realizing the device of various functions either the software module of implementation method can be hardware again Structure in component.

The present invention be referring to according to the method for the embodiment of the present invention, the process of equipment (system) and computer program product Figure and/or block diagram describe.It should be understood that every one stream in flowchart and/or the block diagram can be realized by computer program instructions The combination of process and/or box in journey and/or box and flowchart and/or the block diagram.It can provide these computer programs Instruct the processor of general purpose computer, special purpose computer, Embedded Processor or other programmable data processing devices to produce A raw machine, so that being generated by the instruction that computer or the processor of other programmable data processing devices execute for real The device for the function of being specified in present one or more flows of the flowchart and/or one or more blocks of the block diagram.

These computer program instructions, which may also be stored in, is able to guide computer or other programmable data processing devices with spy Determine in the computer-readable memory that mode works, so that it includes referring to that instruction stored in the computer readable memory, which generates, Enable the manufacture of device, the command device realize in one box of one or more flows of the flowchart and/or block diagram or The function of being specified in multiple boxes.

These computer program instructions also can be loaded onto a computer or other programmable data processing device, so that counting Series of operation steps are executed on calculation machine or other programmable devices to generate computer implemented processing, thus in computer or The instruction executed on other programmable devices is provided for realizing in one or more flows of the flowchart and/or block diagram one The step of function of being specified in a box or multiple boxes.

In a typical configuration, calculating equipment includes one or more processors (CPU), input/output interface, net Network interface and memory.

Memory may include the non-volatile memory in computer-readable medium, random access memory (RAM) and/or The forms such as Nonvolatile memory, such as read-only memory (ROM) or flash memory (flash RAM).Memory is computer-readable medium Example.

Computer-readable medium includes permanent and non-permanent, removable and non-removable media can be by any method Or technology come realize information store.Information can be computer readable instructions, data structure, the module of program or other data. The example of the storage medium of computer includes, but are not limited to phase change memory (PRAM), static random access memory (SRAM), moves State random access memory (DRAM), other kinds of random access memory (RAM), read-only memory (ROM), electric erasable Programmable read only memory (EEPROM), flash memory or other memory techniques, read-only disc read only memory (CD-ROM) (CD-ROM), Digital versatile disc (DVD) or other optical storage, magnetic cassettes, tape magnetic disk storage or other magnetic storage devices Or any other non-transmission medium, can be used for storage can be accessed by a computing device information.As defined in this article, it calculates Machine readable medium does not include temporary computer readable media (transitory media), such as the data-signal and carrier wave of modulation.

It will be understood by those skilled in the art that the embodiment of this specification can provide as the production of method, system or computer program Product.Therefore, in terms of this specification embodiment can be used complete hardware embodiment, complete software embodiment or combine software and hardware Embodiment form.Moreover, it wherein includes computer available programs that this specification embodiment, which can be used in one or more, Implement in the computer-usable storage medium (including but not limited to magnetic disk storage, CD-ROM, optical memory etc.) of code The form of computer program product.

This specification embodiment can describe in the general context of computer-executable instructions executed by a computer, Such as program module.Generally, program module includes routines performing specific tasks or implementing specific abstract data types, journey Sequence, object, component, data structure etc..This specification embodiment can also be practiced in a distributed computing environment, in these points Cloth calculates in environment, by executing task by the connected remote processing devices of communication network.In distributed computing ring In border, program module can be located in the local and remote computer storage media including storage equipment.

All the embodiments in this specification are described in a progressive manner, same and similar portion between each embodiment Dividing may refer to each other, and each embodiment focuses on the differences from other embodiments.Especially for system reality For applying example, since it is substantially similar to the method embodiment, so being described relatively simple, related place is referring to embodiment of the method Part explanation.In the description of this specification, reference term " one embodiment ", " some embodiments ", " example ", The description of " specific example " or " some examples " etc. means specific features described in conjunction with this embodiment or example, structure, material Or feature is contained at least one embodiment or example of this specification embodiment.In the present specification, to above-mentioned term Schematic representation be necessarily directed to identical embodiment or example.Moreover, description specific features, structure, material or Person's feature may be combined in any suitable manner in any one or more of the embodiments or examples.In addition, in not conflicting feelings Under condition, those skilled in the art by different embodiments or examples described in this specification and different embodiment or can show The feature of example is combined.

The foregoing is merely the embodiments of this specification embodiment, are not limited to this specification embodiment.It is right For those skilled in the art, this specification embodiment can have various modifications and variations.It is all in this specification embodiment Any modification, equivalent replacement, improvement and so within spirit and principle, the right that should be included in this specification embodiment are wanted Within the scope of asking.

Claims

1. a kind of finance data risk control method based on artificial intelligence, which is characterized in that the finance data risk control Method includes:

Acquire historical financial data sample, wherein include each Transaction Information and correspondence in the historical financial data sample Type of transaction label；

Based on preset rules, according to each Transaction Information and corresponding type of transaction mark in the historical financial data sample Label, establish finance data air control model；

And the type of transaction of destination financial data is predicted according to the finance data air control model, and according to prediction As a result risk control is carried out to the destination financial data.

2. finance data risk control method according to claim 1, which is characterized in that described to be based on preset rules, root According to each Transaction Information and corresponding type of transaction label in the historical financial data, finance data air control model is established, Include:

The historical financial data sample is pre-processed；

3. finance data risk control method according to claim 2, which is characterized in that described to the historical financial number It is pre-processed according to sample, comprising:

And data preanalysis processing is carried out to through data cleansing treated historical financial data sample, it obtains described through number According to the data structure feature of cleaned historical financial data sample.

4. finance data risk control method according to claim 3, which is characterized in that described by pretreated history Finance data sample is sampled, and obtains training data, comprising:

According to the data structure feature through data cleansing treated historical financial data sample, to described through data cleansing Treated, and historical financial data sample is sampled, and obtains the training data as machine learning；

It wherein, include positive example sample and negative data in the training data, and the positive example sample corresponding first is traded Type label, the corresponding second type of transaction label of the negative data.

5. finance data risk control method according to claim 2, which is characterized in that described to be based on the training data Aspect of model selection is carried out, training characteristics are obtained, comprising:

The training data is pre-processed according to the first preset rules；

6. finance data risk control method according to claim 5, which is characterized in that first preset rules are spy One of the mode of assemble-publish code, feature normalization, feature discretization and Feature Dimension Reduction mode, whole modes or any combination；

Second preset rules are information gain judgment mode and/or the phase relation for the linear relationship between being characterized by Number judgment mode.

7. finance data risk control method according to claim 2, which is characterized in that described according to the training characteristics Model training is carried out to default sorting algorithm, obtains the finance data air control model, comprising:

The input feature vector sequence of the default sorting algorithm is constructed according to each training characteristics and corresponding type of transaction label Column；

And model training is carried out to described pair of default sorting algorithm based on input feature vector sequence, obtain the finance data wind Control model；

8. finance data risk control method according to claim 1, which is characterized in that described according to the finance data Air control model predicts the type of transaction of destination financial data, and carries out wind to the destination financial data according to prediction result Danger control, comprising:

The type of transaction of destination financial data is predicted according to the finance data air control model, prediction obtains the target The corresponding type of transaction label of finance data is the first type of transaction label or the second type of transaction label；

And if the type of transaction label is the first type of transaction label, export corresponding first default risk control knot Fruit；

If the type of transaction label is the second type of transaction label, corresponding second default risk control result is exported.

9. finance data risk control method according to any one of claims 1 to 8, which is characterized in that in the acquisition Before historical financial data sample, the finance data risk control method further include:

And the corresponding type of transaction label of each Transaction Information is generated according to the corresponding type of transaction of each process of exchange；

Each Transaction Information is stored into historical financial data sample database, wherein the historical financial data sample The corresponding relationship being stored in database between each Transaction Information and type of transaction label.

10. finance data risk control method according to claim 9, which is characterized in that the historical financial data sample Database is distributed data base, and the distributed data base is stored in block chain network.

11. a kind of finance data risk control system based on artificial intelligence, which is characterized in that the finance data risk control System includes:

Historical financial data sample collection module, for acquiring historical financial data sample, wherein the historical financial data sample It include each Transaction Information and corresponding type of transaction label in this；

Finance data air control model construction module, for based on preset rules, according to each in the historical financial data sample A Transaction Information and corresponding type of transaction label, establish finance data air control model；

Risk control module, it is pre- for being carried out according to type of transaction of the finance data air control model to destination financial data It surveys, and risk control is carried out to the destination financial data according to prediction result.

12. a kind of electronic equipment including memory, processor and stores the calculating that can be run on a memory and on a processor Machine program, which is characterized in that the processor realizes that claims 1 to 10 is described in any item and is based on people when executing described program The step of finance data risk control method of work intelligence.

13. a kind of computer readable storage medium, is stored thereon with computer program, which is characterized in that the computer program quilt Claims 1 to 10 described in any item finance data risk control methods based on artificial intelligence are realized when processor executes Step.