CN110209729A - The method and device of data transfer object identification - Google Patents
The method and device of data transfer object identification Download PDFInfo
- Publication number
- CN110209729A CN110209729A CN201910323746.9A CN201910323746A CN110209729A CN 110209729 A CN110209729 A CN 110209729A CN 201910323746 A CN201910323746 A CN 201910323746A CN 110209729 A CN110209729 A CN 110209729A
- Authority
- CN
- China
- Prior art keywords
- data transfer
- transfer object
- data
- stream compression
- statistical nature
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/27—Replication, distribution or synchronisation of data between databases or within a distributed database system; Distributed database system architectures therefor
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Computing Systems (AREA)
- Databases & Information Systems (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Compression, Expansion, Code Conversion, And Decoders (AREA)
Abstract
The embodiment of the present application discloses a kind of method and device of data transfer object identification, includes the stream compression managing detailed catalogue between at least two data transfer objects in stream compression information this method comprises: obtaining the stream compression information of data transfer object;The stream compression relationship between at least two data transfer objects is constructed based on stream compression managing detailed catalogue, and extracts the data transfer object statistical nature of each data transfer object in at least two data transfer objects for including in stream compression managing detailed catalogue;Based on picture scroll product neural network model, the data transfer object statistical nature of each data transfer object, determines the data transfer object classification of each data transfer object at least two data transfer objects in combined data circulation relationship and at least two data transfer objects.Using the embodiment of the present application, the classification of data transfer object can be quickly recognized, improves the accuracy rate of data transfer object identification, applicability is high.
Description
Technical field
This application involves the method and devices that field of computer technology more particularly to a kind of data transfer object identify.
Background technique
With the rise of mobile internet, for reinforce each data transfer object (such as data storage card, SIM card or
Bank card etc.) between data exchange, the data between multiple data transfer objects mutually shift or data mutually circulate and have become
For the common method of Data Migration or data sharing.Currently, being able to carry out the quantity right and wrong of the data transfer object of data circulation
It is often huge, and the type of data transfer object is very different, especially in real data transfer process, if user is not knowing
In the case that whether data transfer object is legal, some key/critical data have been transferred in invalid data transfer object by mistake, then
It is very likely to cause the heavy losses of user or enterprise.Therefore how quickly and accurately to identify invalid data transfer object and
The stream compression of effectively control invalid data transfer object becomes current urgent problem to be solved.
Summary of the invention
The embodiment of the present application provides a kind of method and device of data transfer object identification.Data transfer can be quickly recognized
The classification of object, improves the accuracy rate of data transfer object identification, and applicability is high.
In a first aspect, the embodiment of the present application provides a kind of data transfer object knowledge method for distinguishing, this method comprises:
The stream compression information of data transfer object is obtained, includes that at least two data shift in above-mentioned stream compression information
Stream compression managing detailed catalogue between object;
The stream compression between above-mentioned at least two data transfer object is constructed based on above-mentioned stream compression managing detailed catalogue to close
System, and extract each data transfer pair in the above-mentioned at least two data transfer object for including in above-mentioned stream compression managing detailed catalogue
The data transfer object statistical nature of elephant;
Based on picture scroll product neural network model, in conjunction with above-mentioned stream compression relationship and above-mentioned at least two data transfer object
In each data transfer object data transfer object statistical nature, determine in above-mentioned at least two data transfer object each
The data transfer object classification of data transfer object.
The embodiment of the present application can be based on picture scroll product neural network model, in conjunction with above-mentioned stream compression relationship and above-mentioned at least two
The data transfer object statistical nature of each data transfer object quickly and accurately identifies each in a data transfer object
The data transfer object classification of data transfer object.
With reference to first aspect, above-mentioned to be constructed based on above-mentioned stream compression managing detailed catalogue in a kind of possible embodiment
Stream compression relationship between above-mentioned at least two data transfer object, comprising:
Extract the pipelined data in above-mentioned stream compression managing detailed catalogue;
According to the data transfer relationship for each data transfer object for including in above-mentioned pipelined data, building above-mentioned at least two
Stream compression relationship between data transfer object;
Wherein, including at least data are transferred to, data produce, the data side of being transferred to and data connect in above-mentioned data transfer relationship
Debit.
The embodiment of the present application can construct the stream compression relationship between each data transfer object, improve the data flow of building
The integrality and accuracy transferred the registration of Party membership, etc. from one unit to another.
With reference to first aspect, above-mentioned based on picture scroll product neural network model in a kind of possible embodiment, in conjunction with upper
State the data transfer object statistics of each data transfer object in stream compression relationship and above-mentioned at least two data transfer object
Feature determines the data transfer object classification of each data transfer object in above-mentioned at least two data transfer object, comprising:
The corresponding neighbour of stream compression between above-mentioned at least two data transfer object is constructed according to above-mentioned stream compression relationship
Connect matrix;
According to the data transfer object statistical nature of data transfer object each in above-mentioned at least two data transfer object
Construct data transfer object statistical nature matrix, wherein the data transfer object statistical nature of above-mentioned each data transfer object
Number of days, average data surplus, associated data transfer object number and average data are enlivened including stream compression number of days, stream compression
At least one of in transfer amount;
By above-mentioned adjacency matrix and above-mentioned data transfer object statistical nature Input matrix picture scroll product neural network model, and
Output result based on above-mentioned picture scroll product neural network model determines each data in above-mentioned at least two data transfer object
The data transfer object classification of transfer object.
The embodiment of the present application passes through the input matrix of structure figures convolutional neural networks model, standardizes format more, and number
According to the statistical nature in transfer object statistical nature matrix including multiple interdependent nodes, figure convolutional neural networks are helped to improve
The accuracy of model output.
With reference to first aspect, above-mentioned according to above-mentioned at least two data transfer object in a kind of possible embodiment
In each data transfer object data transfer object statistical nature construct data transfer object statistical nature matrix, comprising:
By the data transfer object statistical nature of data transfer object each in above-mentioned at least two data transfer object point
It is not normalized to obtain the corresponding normalization of each data transfer object in above-mentioned at least two data transfer object
Treated data transfer object statistical nature;
Based on the number after the corresponding normalized of data transfer object each in above-mentioned at least two data transfer object
Data transfer object statistical nature matrix is constructed according to transfer object statistical nature.
The embodiment of the present application eliminates the dimension impact of different statistical natures, and then improves picture scroll product neural network model
Export the accuracy and reliability of result.
With reference to first aspect, in a kind of possible embodiment, the above method further include:
Training sample is obtained, includes the stream compression detail between multiple sample data transfer objects in above-mentioned training sample
The data transfer object classification of information and each sample data transfer object;
The multiple sample number is constructed according to the stream compression managing detailed catalogue between above-mentioned multiple sample data transfer objects
According to the corresponding adjacency matrix of stream compression between transfer object and data transfer object statistical nature matrix;
The corresponding adjacency matrix of stream compression managing detailed catalogue, data transfer based on above-mentioned multiple sample data transfer objects
The data transfer object classification of object statistical nature matrix and above-mentioned each sample data transfer object, structure figures convolutional Neural net
Network model, so that above-mentioned picture scroll product neural network model has according to the corresponding adjoining of any data of input circulation managing detailed catalogue
The data transfer pair for including in matrix and any of the above-described stream compression managing detailed catalogue of data transfer object statistical nature Output matrix
The ability of the data transfer object classification of elephant.
The embodiment of the present application can train to obtain the ability for the data transfer object classification for having identification data transfer object
Picture scroll accumulates neural network model, and then improves the recognition speed to data transfer object classification and enhance the accurate of recognition result
Property.
With reference to first aspect, in a kind of possible embodiment, the above method further include:
If the data transfer object classification of any data transfer object is target in above-mentioned at least two data transfer object
Data transfer object classification then sends data transfer object stream compression abnormality alarming information to above-mentioned any data transfer object
Data transfer object manager to remind above-mentioned data transfer object manager to the data of any of the above-described data transfer object
Circulation is supervised.
The embodiment of the present application can be reinforced by sending a warning message to data transfer object manager to different classes of number
According to the supervision of transfer object.
With reference to first aspect, in a kind of possible embodiment, the above method further include:
If the data transfer object classification of any data transfer object is target in above-mentioned at least two data transfer object
Data transfer object classification locks any of the above-described data transfer object then to block the data flow of any of the above-described data transfer object
Turn.
The embodiment of the present application is by blocking the stream compression of the invalid data transfer object identified that can stop loss in time.
Second aspect, the embodiment of the present application provide a kind of device of data transfer object identification, which includes:
Stream compression data obtaining module, for obtaining the stream compression information of data transfer object, above-mentioned stream compression
It include the stream compression managing detailed catalogue between at least two data transfer objects in information;
Data transfer object characteristic determination module, the above-mentioned number for being determined based on above-mentioned stream compression data obtaining module
The stream compression relationship between above-mentioned at least two data transfer object is constructed according to circulation managing detailed catalogue, and extracts above-mentioned data flow
Turn the data transfer object system of each data transfer object in the above-mentioned at least two data transfer object for including in managing detailed catalogue
Count feature;
Data transfer object identification module, for being based on picture scroll product neural network model, in conjunction with above-mentioned data transfer object
Each data transfer pair in the above-mentioned stream compression relationship and above-mentioned at least two data transfer object that characteristic determination module determines
The data transfer object statistical nature of elephant, determines the number of each data transfer object in above-mentioned at least two data transfer object
According to transfer object classification.
In conjunction with second aspect, in a kind of possible embodiment, above-mentioned data transfer object characteristic determination module includes:
Pipelined data acquiring unit, for extracting the pipelined data in above-mentioned stream compression managing detailed catalogue;
Stream compression relationship construction unit, for the data according to each data transfer object for including in above-mentioned pipelined data
Transfer relationship constructs the stream compression relationship between above-mentioned at least two data transfer object;
Wherein, including at least data are transferred to, data produce, the data side of being transferred to and data connect in above-mentioned data transfer relationship
Debit.
In conjunction with second aspect, in a kind of possible embodiment, above-mentioned data transfer object identification module includes:
Adjacency matrix construction unit, for constructing above-mentioned at least two data transfer object according to above-mentioned stream compression relationship
Between the corresponding adjacency matrix of stream compression;
Data transfer object statistical nature matrix unit, for according to every number in above-mentioned at least two data transfer object
Data transfer object statistical nature matrix is constructed according to the data transfer object statistical nature of transfer object, wherein above-mentioned every number
Data transfer object statistical nature according to transfer object includes that stream compression number of days, stream compression enliven number of days, more than average data
At least one of in amount, associated data transfer object number and average data transfer amount;
Data transfer object classification determination unit is used for above-mentioned adjacency matrix and above-mentioned data transfer object statistical nature
Input matrix picture scroll accumulate neural network model, and based on above-mentioned picture scroll product neural network model output result determine it is above-mentioned extremely
The data transfer object classification of each data transfer object in few two data transfer objects.
In conjunction with second aspect, in a kind of possible embodiment, above-mentioned data transfer object statistical nature matrix unit
For:
By the data transfer object statistical nature of data transfer object each in above-mentioned at least two data transfer object point
It is not normalized to obtain the corresponding normalization of each data transfer object in above-mentioned at least two data transfer object
Treated data transfer object statistical nature;
Based on the number after the corresponding normalized of data transfer object each in above-mentioned at least two data transfer object
Data transfer object statistical nature matrix is constructed according to transfer object statistical nature.
In conjunction with second aspect, in a kind of possible embodiment, above-mentioned data transfer object identification device further includes figure
Convolutional neural networks model determining module, wherein above-mentioned picture scroll accumulates neural network model determining module and includes:
Training sample data acquiring unit includes multiple sample datas in above-mentioned training sample for obtaining training sample
The data transfer object classification of stream compression managing detailed catalogue and each sample data transfer object between transfer object;
Training sample data processing unit, for bright according to the stream compression between above-mentioned multiple sample data transfer objects
The corresponding adjacency matrix of stream compression and data transfer object between the multiple sample data transfer object of thin information architecture
Statistical nature matrix;
Picture scroll accumulates neural network model training unit, for the stream compression based on above-mentioned multiple sample data transfer objects
The number of the corresponding adjacency matrix of managing detailed catalogue, data transfer object statistical nature matrix and above-mentioned each sample data transfer object
According to transfer object classification, structure figures convolutional neural networks model, so that above-mentioned picture scroll product neural network model has according to input
Any data corresponding adjacency matrix of circulation managing detailed catalogue and any of the above-described number of data transfer object statistical nature Output matrix
According to the ability of the data transfer object classification for the data transfer object for including in circulation managing detailed catalogue.
In conjunction with second aspect, in a kind of possible embodiment, above-mentioned data transfer object identification device further include:
The first administration module of data transfer object, if for any data transfer in above-mentioned at least two data transfer object
The data transfer object classification of object is target data transfer object classification, then sends data transfer object stream compression and accuse extremely
Alert information reminds above-mentioned data transfer object manager to the data transfer object manager of above-mentioned any data transfer object
The stream compression of any of the above-described data transfer object is supervised.
In conjunction with second aspect, in a kind of possible embodiment, above-mentioned data transfer object identification device further include:
The second administration module of data transfer object, if for any data transfer in above-mentioned at least two data transfer object
The data transfer object classification of object is target data transfer object classification, then locks any of the above-described data transfer object to block
The stream compression of any of the above-described data transfer object.
The third aspect, the embodiment of the present application provide a kind of terminal device, which includes processor and memory,
The processor and memory are connected with each other.The memory for store support the terminal device execute above-mentioned first aspect and/or
The computer program for the method that any possible implementation of first aspect provides, which includes program instruction,
The processor is configured for calling above procedure instruction, executes above-mentioned first aspect and/or first aspect is any possible
Method provided by embodiment.
Fourth aspect, the embodiment of the present application provide a kind of computer readable storage medium, the computer-readable storage medium
Matter is stored with computer program, which includes program instruction, which makes at this when being executed by a processor
It manages device and executes method provided by above-mentioned first aspect and/or any possible embodiment of first aspect.
Implement the embodiment of the present application, has the following beneficial effects:
Picture scroll product neural network model based on building, and combine above-mentioned stream compression relationship and above-mentioned at least two data
The data transfer object statistical nature of each data transfer object, can quickly and accurately identify each data in transfer object
The data transfer object classification of transfer object.
Detailed description of the invention
In order to more clearly explain the technical solutions in the embodiments of the present application, make required in being described below to embodiment
Attached drawing is briefly described, it should be apparent that, the drawings in the following description are only some examples of the present application, for
For those of ordinary skill in the art, without creative efforts, it can also be obtained according to these attached drawings other
Attached drawing.
Fig. 1 is a flow diagram of data transfer object recognition methods provided by the embodiments of the present application;
Fig. 2 a is an application scenarios schematic diagram of the stream compression information of data transfer object provided by the embodiments of the present application;
Fig. 2 b is the another application scene signal of the stream compression information of data transfer object provided by the embodiments of the present application
Figure;
Fig. 3 is the schematic diagram of once transfer relationship provided by the embodiments of the present application and two degree of transfer relationships;
Fig. 4 is the application scenarios schematic diagram of stream compression managing detailed catalogue provided by the embodiments of the present application;
Fig. 5 is building stream compression relationship provided by the embodiments of the present application and extracts data transfer object statistical nature and answer
Use schematic diagram of a scenario;
Fig. 6 is the schematic diagram of building adjacency matrix provided by the embodiments of the present application;
Fig. 7 is the schematic diagram of building data transfer object statistical nature matrix provided by the embodiments of the present application;
Fig. 8 is the schematic diagram of picture scroll product neural network model provided by the embodiments of the present application;
Fig. 9 is showing based on picture scroll product neural network model identification data transfer object classification provided by the embodiments of the present application
It is intended to;
Figure 10 is matrix building schematic diagram provided by the embodiments of the present application;
Figure 11 is another flow diagram of data transfer object recognition methods provided by the embodiments of the present application;
Figure 12 is the structural schematic diagram of data transfer object identification device provided by the embodiments of the present application;
Figure 13 is the structural schematic diagram of terminal device provided by the embodiments of the present application.
Specific embodiment
Below in conjunction with the attached drawing in the embodiment of the present application, technical solutions in the embodiments of the present application carries out clear, complete
Site preparation description, it is clear that described embodiments are only a part of embodiments of the present application, instead of all the embodiments.It is based on
Embodiment in the application, it is obtained by those of ordinary skill in the art without making creative efforts every other
Embodiment shall fall in the protection scope of this application.
Data transfer object provided by the embodiments of the present application knows method for distinguishing, is widely portable to various be able to carry out data
The identifying system or terminal device of the object (including data storage card, stored value card, bank card or SIM card etc.) of transfer.Wherein,
Above-mentioned terminal device includes but is not limited to smart phone, tablet computer, laptop and desktop computer etc., is not limited herein
System.For convenience of description, will be illustrated by taking terminal device as an example below.
Method provided by the embodiments of the present application can be based on data flow by the stream compression information of acquisition data transfer object
The stream compression managing detailed catalogue between at least two data transfer objects for including in transfering the letter breath, building at least two data transfer
Stream compression relationship between object, and extract the data transfer of each data transfer object at least two data transfer objects
Object statistical nature, in conjunction with picture scroll product neural network model, it may be determined that go out each in at least two data transfer object
The data transfer object classification of data transfer object.Using method provided by the embodiments of the present application, data can be quickly recognized and turned
The classification of object is moved, the accuracy rate of data transfer object identification is improved, applicability is high.
Method provided by the embodiments of the present application and relevant apparatus are carried out respectively respectively below in conjunction with Fig. 1 to Figure 13 in detail
Explanation.It may include for obtaining stream compression information, creation stream compression relationship, extracting in method provided by the embodiments of the present application
Data transfer object statistical nature and the data transfer that each data transfer object is identified based on picture scroll product neural network model
The data processing stages such as object type.Wherein, the implementation of above-mentioned each data processing stage can be found in following Fig. 1 and Figure 11
Shown in implementation.
It is a flow diagram of data transfer object recognition methods provided by the embodiments of the present application referring to Fig. 1, Fig. 1.This
The method that application embodiment provides may include steps of 101 to 103:
101, the stream compression information of data transfer object is obtained.
In some possible embodiments, it can be generated in operation with the network equipment and record network equipment operation
The journal file of situation is similar, and any data transfer object can also generate corresponding record data transfer pair in actual operation
The data transfer process state of elephant and the stream compression information of data transfer process result.Here, data transfer object can be
Bank card or SIM card etc., herein with no restrictions.For example, any bank card, can generate a series of in user's use process
Transaction journal information, these transaction journal information are able to reflect the basic condition of passing business transaction, therefore these transaction flows
Water information can serve as the stream compression information of the bank card.A referring to fig. 2, Fig. 2 a are data provided by the embodiments of the present application
One application scenarios schematic diagram of the stream compression information of transfer object, in fig. 2 a, data transfer object are bank card, data flow
It may include card number, trade date, loco, transaction summarization, currency type, transaction amount, account balance, reciprocal account in transfering the letter breath
And transaction teller number etc., wherein transaction summarization includes transferring accounts, depositing, withdrawing the money, consuming, wage, service charge, settling interests and manage money matters
Deng herein with no restrictions.In another example user using SIM card when being made a phone call or being sent short messages on mobile phone, it will usually generate a series of
Message registration or short message record, these message registrations or short message record can also be used as the stream compression information of SIM card.Referring to
Fig. 2 b, Fig. 2 b are the another application schematic diagram of a scenario of the stream compression information of data transfer object provided by the embodiments of the present application,
In figure 2b, data transfer object is SIM card, may include telephone number, call date, operator, ditch in stream compression information
Logical abstract, number attribution, account balance, other party number and the duration of call/number etc., wherein link up abstract including caller,
It is called, supplements with money, short message, access failure, rejection etc., herein with no restrictions.
In some possible embodiments, can by being docked with the data system of each bank or operator, with
The permission for entering the data system of each bank or operator is obtained, and then can be from the data system of each bank or operator
The stream compression information of one or more specified data transfer object (such as bank card or SIM card) is obtained as to be identified
The stream compression information of data transfer object wherein includes the number between at least two data transfer objects in stream compression information
According to circulation managing detailed catalogue.Optionally, can also continue to obtain after getting the stream compression information of some data transfer object
Take the stream compression information for other data transfer objects for having N degree transfer relationship with the data transfer object.Here, N is greater than 0
Integer, for example, N degree transfer relationship can be once transfer relationship, two degree of transfer relationships or three degree of transfer relationships etc., herein not
It is limited.It is the schematic diagram of once transfer relationship provided by the embodiments of the present application and two degree of transfer relationships referring to Fig. 3, Fig. 3, such as schemes
Shown in 3, it is assumed that one data transfer object of each node on behalf in figure, by taking the stream compression relationship of data transfer object A1 as an example
Once transfer relationship is explained, then being to shift with data with the data transfer object A1 data transfer object for having once transfer relationship
Object A1 has data transfer object B1, the data transfer object C1 and data transfer object D1 of immediate data circulation relationship.Wherein
With the once transfer relationship that the data transfer object A1 data transfer object for having two degree of transfer relationships is with data transfer object A1
Data transfer object have immediate data circulation relationship data transfer object, such as data transfer object E1, the number for including in Fig. 3
According to transfer object F1, data transfer object G1 and data transfer object H1.In the embodiment of the present application, our usual taking N
Value is set as 2, i.e., after the stream compression information for obtaining a certain data transfer object, can also continue to obtain and the data turn
Shifting object has the stream compression information of the data transfer object of once transfer relationship and two degree of transfer relationships as data to be identified
The stream compression information of transfer object wherein includes the data flow between at least two data transfer objects in stream compression information
Turn managing detailed catalogue.
102, the stream compression relationship between at least two data transfer objects is constructed based on stream compression managing detailed catalogue, and
Extract the data transfer of each data transfer object in at least two data transfer objects for including in stream compression managing detailed catalogue
Object statistical nature.
In some possible embodiments, in the stream compression information of acquisition include at least two data transfer objects it
Between stream compression managing detailed catalogue, for example, including at least in stream compression information so that data transfer object is bank card as an example
Stream compression managing detailed catalogue between two data transfer objects can be the managing detailed catalogue of transferring accounts between at least two bank cards.Again
For example, it is assumed that data transfer object is SIM card, the number between at least two data transfer objects for including in stream compression information
According to the message registration that circulation managing detailed catalogue can be between at least two SIM cards.Further, by extracting each data transfer object
Stream compression managing detailed catalogue in pipelined data, and according to the data for each data transfer object for including in pipelined data shift
Relationship can construct the stream compression relationship between at least two data transfer objects, while can also extract stream compression detail
The data transfer object statistical nature of each data transfer object in at least two data transfer objects for including in information.This
In, it may include data transfer relationship, data transfer time, data transfer amount and data surplus etc., data transfer in pipelined data
Include at least that data are transferred to, data produce, the data side of being transferred to and data receiver in relationship.It is understandable to be, a number
It include that data are transferred to and produce with the data side of being transferred to or data according to the data transfer relationship in any bar pipelined data of transfer object
And data receiver.Referring to fig. 4, Fig. 4 is the application scenarios schematic diagram of stream compression managing detailed catalogue provided by the embodiments of the present application.
The data transfer relationship that data transfer object (i.e. bank card) is explained by taking first pipelined data of bank card in Fig. 4 as an example,
In the data transfer relationship of bank card 11***11, according to transaction amount 2600 (transaction amount is positive) and reciprocal account 77***77
It is found that data transfer relationship is that data are transferred to and the data side of being transferred to is that (i.e. bank card 77***77 is to bank by bank card 77***77
Card 11***11 is transferred to 2600 yuan).In another example, it is assumed that data transfer object is SIM card, with a message registration of certain SIM card
The data transfer relationship of data transfer object is explained for information, it is assumed that the communication abstract of its message registration is caller, then can be true
The data transfer relationship for making the SIM card is that data produce and the data side of producing is the other party number in message registration.Wherein, often
The data transfer object statistical nature of a data transfer object may include that stream compression number of days, stream compression enliven number of days, average
Data surplus, associated data transfer object number and Partial Feature or whole features in average data transfer amount, do not do herein
Limitation.Understandable to be, stream compression number of days is that some data transfer object has stream compression information note within certain time
The number of days of record, it is that some data transfer object data in the stream compression information in certain time turn that stream compression, which enlivens number of days,
Shifting amount is greater than the number of days of preset data transfer amount, it is however generally that, stream compression enlivens number of days less than or equal to stream compression day
Number.Average data surplus is the arithmetic mean of instantaneous value of some data transfer object the sum of data surplus within certain time, for example, false
If the data surplus for having recorded first day in a certain nearest three days stream compression information of data transfer object is x, second day
Data surplus is y, and the data surplus in third day is z, then nearest three days average data surpluses of the data transfer object are (x+y+
z)/3.Associated data transfer object number is to have data transfer relationship with some data transfer object (data are transferred to be produced with data)
Data transfer object quantity.For example, the associated data transfer object number of data transfer object A1 is 3 in figure by taking Fig. 3 as an example
(being data transfer object B1, C1 and the D1 for thering are data to produce relationship with data transfer object A1 respectively), data transfer object C1
Associated data transfer object number be 3 (be the data transfer object A1 for thering are data to be transferred to relationship with data transfer object C1 respectively
With data the transfer object F1 and G1 for thering are data to produce relationship with data transfer object C1).Average data transfer amount includes average
It is transferred to data transfer amount and averagely produces data transfer amount, wherein be averagely transferred to data transfer amount=data and be transferred to transfer total amount
÷ is transferred to number, averagely produce data transfer amount=data produce transfer total amount ÷ produce number.For example, by taking A1 in Fig. 3 as an example,
Assuming that having recorded data transfer object A1 in nearest three days stream compression information of data transfer object A1 has produced data transfer
Amount a gives data transfer object B1, data transfer object A1 to produce data transfer amount b and gives data transfer object C1, data transfer
Object A1 has produced data transfer amount c to data transfer object D1 and data transfer object A1 does not receive any data transfer pair
As the data being transferred to, then data transfer object A1 is averaged that produce data transfer amount be (a+b+c)/3, data transfer object A1's
Averagely being transferred to data transfer amount is 0.
In some possible embodiments, the stream compression relationship of building, which can be, can embody each data transfer pair
The topological diagram of data transfer relationship as between or all data transfer relationships that can be embodied between each data transfer object
Other forms, such as table.It is building stream compression provided by the embodiments of the present application referring to Fig. 5, Fig. 5 by taking bank card as an example
Relationship and the application scenarios schematic diagram for extracting data transfer object statistical nature.As shown in figure 5, with first of bank card in Fig. 5
The implementation of building stream compression relationship is illustrated for pipelined data.It is shifted in the data of bank card 11***11
In relationship, according to transaction amount 2600 (transaction amount is positive) and reciprocal account 77***77, bank card in topological diagram can be constructed
A data transfer relationship of 11***11 and bank card 77***77 is bank card 77***77 → bank card 11***11.With such
It pushes away, the transaction journal data in managing detailed catalogue of transferring accounts (i.e. stream compression managing detailed catalogue) by extracting each bank card (flow
Water number evidence), it can be constructed each according to the relationship of transferring accounts (i.e. data transfer relationship) between each bank card for including in transaction journal data
Open the topological diagram (i.e. stream compression relationship) for behavior of transferring accounts between bank card.It here, may include pass of transferring accounts in transaction journal data
System, trade date, transaction amount and account balance etc., wherein the relationship of transferring accounts may include that data are transferred to (transaction amount is positive), number
It (trades according to (transaction amount is negative), the data side of being transferred to (reciprocal account that transaction amount is timing) and data receiver is produced
The reciprocal account when amount of money is negative).The data transfer object for each data transfer object (every bank card) wherein extracted is united
Meter feature (the feature x11 ... feature x1n of such as bank card 11***11) includes stream compression number of days (transaction number of days), stream compression
It enlivens number of days (enlivening number of days), average data surplus (average balance), associated data transfer object number (opponent's card number) and puts down
Equal data transfer amount (average transaction amount) etc..In another example, it is assumed that data transfer object is SIM card, by extracting each SIM card
Message registration information (i.e. stream compression managing detailed catalogue) in communicating data (i.e. pipelined data), can be wrapped according in communicating data
Calling relationship (i.e. data transfer relationship) between each SIM card included constructs and calls the topological diagram of behavior (i.e. between each SIM card
Stream compression relationship).Here, when may include calling relationship, calling date, the number of calls, connection number, call in communicating data
It is long etc., wherein calling relationship may include that data are transferred to (called), data produce (caller), the data side of being transferred to (other side when called
Number) and data receiver's (other party number when caller).The each data transfer object (every SIM card) wherein extracted
Data transfer object statistical nature (SIM card feature) includes that stream compression number of days (calling number of days), stream compression enliven number of days and (exhale
Cry and enliven number of days), average data surplus (average call duration), associated data transfer object number (association SIM card number) and flat
Equal data transfer amount (average call number) etc..
103, the data transfer object classification of each data transfer object is identified based on picture scroll product neural network model.
Since 2012, deep learning achieves huge in two fields of computer vision and natural language processing
Success.The appearance of especially convolutional neural networks enables some challenges preferably to be solved, but convolutional Neural
The research object of network is that have the data of the space structure of rule, for example voice is the one-dimensional sequence of rule, for another example picture
It is the square grid of rule, by the data of these space structures for having rule by being indicated with one-dimensional or two-dimensional matrix
Afterwards, convolutional neural networks is recycled to deal with just very efficiently.But there are many data not have in our actual life
What the space structure, such as recommender system, electronic trading system, computational geometry, brain signal, molecular structure etc. of standby rule took out
Map.Each node connection is not quite similar in these map structures, and some nodes are there are three connection, and there are two connect some nodes
It connects, is irregular data structure.Therefore, a kind of method that can carry out deep learning to diagram data is come into being, i.e. picture scroll product
Neural network (Graph Convolutional Network, GCN).It is understood that figure (graph) is a kind of data lattice
Formula, it can be used to indicate that social networks, communication network, protein molecular network etc., it is however generally that the node in figure is for indicating
Individual in network, the side in figure are used to indicate the connection relationship in network between individual.Therefore one can consider that diagram data
Have two characteristics, first is that each node has the characteristic information of oneself, second is that each node in figure also has structural information.
Wherein GCN model then can be based on the node diagnostic information and node structure information progress node-classification in the diagram data of input
Or side prediction.In the embodiment of the present application, based on picture scroll product neural network model, in conjunction with the data between each data transfer object
Data transfer object statistical nature (the node diagnostic letter of circulation relationship (i.e. node structure information) and each data transfer object
Breath), it may be determined that go out the data transfer object classification of each data transfer object.
Specifically, according to the stream compression relationship (such as topological diagram) of building, it can be by data each in stream compression relationship
Data transfer relationship (neighbouring relations i.e. in topological diagram between vertex) between transfer object is built into adjacency matrix A, citing
For, it is the schematic diagram of building adjacency matrix provided by the embodiments of the present application referring to Fig. 6, Fig. 6.As shown in fig. 6, adjacency matrix A
The direct reachability relation in topological diagram between each node (each data transfer object) is illustrated, in adjacency matrix A
For a line [0 00000 1], the first row in adjacency matrix A illustrates that data transfer object 1 is shifted with data respectively
Object 1, data transfer object 2, data transfer object 3, data transfer object 4, data transfer object 5,6 and of data transfer object
Direct reachability relation between data transfer object 7, wherein without direct reachability relation, 1 table between 0 expression data transfer object
Registration is according to there is direct reachability relation between transfer object.Then, count special according to the data transfer object of each data transfer object
Sign, is built into a data transfer object statistical nature matrix for the data transfer object statistical nature of each data transfer object
X, wherein the data transfer object statistical nature of each data transfer object includes that stream compression number of days, stream compression enliven day
Number, average data surplus, associated data transfer object number and average data transfer amount etc..It is the application reality referring to Fig. 7, Fig. 7
The schematic diagram of the building data transfer object statistical nature matrix of example offer is provided.As shown in fig. 7, data transfer object statistical nature
Matrix X illustrates the corresponding each number of each node (each data transfer object) extracted from stream compression managing detailed catalogue
According to transfer object statistical nature (i.e. F1 ... Fn), with the first row [x in data transfer object statistical nature matrix X11 ...
x1n] for, the first row in data transfer object statistical nature matrix X illustrates the data that data transfer object 1 respectively includes
Transfer object statistical nature x11 ... feature x1n, wherein each data transfer object statistics that any data transfer object includes
Feature includes that stream compression number of days, stream compression enliven number of days, average data surplus, associated data transfer object number and average
Data transfer amount etc..Pass through the adjacency matrix A and data transfer object statistical nature matrix X input figure convolutional Neural that will be constructed
Network model can determine the data transfer pair of each data transfer object based on the output result of picture scroll product neural network model
As classification.It is the schematic diagram of picture scroll product neural network model provided by the embodiments of the present application referring to Fig. 8, Fig. 8.The embodiment of the present application
In picture scroll product neural network model include input layer I, the first picture scroll lamination G1, the second picture scroll lamination G2, depth layer D3 and defeated
Layer O out, by the input that each node intermediate node structural information and node diagnostic information input picture scroll are accumulated to neural network model
Layer I, by input layer I to the mapping of the first convolutional layer G1, mapping of the first convolutional layer G1 to the second convolutional layer G2, the second convolution
It, can be in the defeated of picture scroll product neural network model after the mapping of layer G2 to depth layer D3 and the mapping of depth layer D3 to output layer O
Layer O obtains the node classification of each node out.It is provided by the embodiments of the present application based on figure convolutional neural networks referring to Fig. 9, Fig. 9
The schematic diagram of model identification data transfer object classification.That is the stream compression information of the data transfer object based on acquisition, can structure
Build the data transfer object of the stream compression relationship (node structure information) and each data transfer object between data transfer object
Statistical nature (node diagnostic information), then stream compression relationship and data transfer object statistical nature are by picture scroll product nerve net
Mapping layer by layer in network model, it may be determined that go out the data transfer object classification of each data transfer object.The embodiment of the present application can
The data transfer object classification identified mainly includes two classes, and one kind is valid data transfer object, and another kind of is invalid data
Transfer object.For example, one kind is normal bank card for bank card, it is another kind of be for money laundering illegal bank card or
The abnormal bank cards such as person transfers accounts frequently, amount of transferring accounts is big.In another example one kind is normal SIM card, another kind of for SIM card
It is businessman or the illegal SIM card that swindling gang is used to market or swindle, or the harassing and wrecking for purposes such as product distribution, advertisements
SIM card etc..
Specifically, picture scroll product neural network model can be defined as:
Whereinα is a hyper parameter, for balance own node feature with
The ratio of neighbor node feature, D are the degree matrix generated according to adjacency matrix A, and J is the unit square of all nodes in topological diagram
Battle array, W0∈RI×G1Indicate the weight matrix between input layer I to the first picture scroll lamination G1, effect is to be mapped to node diagnostic
First layer hidden layer.W1∈RG1×G2Indicate the weight matrix between picture scroll lamination G1 to picture scroll lamination G2, effect is by first
The information MAP of layer hidden layer is to second layer hidden layer.W2∈RG2×D3Indicate the weight matrix of picture scroll lamination G2 to depth layer D3,
Its effect is by the information MAP of second layer hidden layer to third layer hidden layer.W3∈RD3×OIndicate depth layer D3 to output layer O
Weight matrix, effect be by the information MAP of third layer hidden layer to output layer O, then the expression of each node is passed throughThe prediction probability of each node is obtained, wherein n is the corresponding number of tags of node
Amount.In the embodiment of the present application, n is data transfer object categorical measure, for example, the value of n can be 2, weight matrix W0~W3
It can be by being obtained after being trained to picture scroll product neural network model.Output result based on picture scroll product neural network model is (each
The prediction probability of a node) it can determine that the data transfer object classification of data transfer object representated by each node.For example,
It is matrix building schematic diagram provided by the embodiments of the present application referring to Figure 10, Figure 10, in Figure 10, degree matrix D is according to adjoining
The diagonal matrix that the degree for the expression node that matrix A generates is constituted, unit matrix J is the unit matrix of data transfer object.
Optionally, in depth learning technology field, different evaluation index is (i.e. in data transfer object statistical nature not
It is exactly the different evaluation index with feature) often there is different dimension and dimensional unit, such situation influences whether
Data analysis as a result, for example by taking data transfer object is bank card as an example, opponent's card number of a bank card is generally several,
And average transaction amount is generally several hundred members or thousands of members or tens of thousands of members, in order to eliminate the dimension impact between each feature, needs
Data normalization processing is carried out, to solve the comparativity between data target.In other words, by being counted to initial data
According to standardization, so that each index is in the same order of magnitude, so as to subsequent progress Comprehensive Correlation evaluation.Wherein, most typical
Data standardization processing method is exactly the normalized of data, and the data after normalized can be limited at centainly
In the range of (such as [0,1] or [- 1,1]).In the embodiment of the present application, range transformation method or the standardization of 0 mean value can be used
Method is active to stream compression number of days included in the data transfer object statistical nature of any data transfer object, stream compression
The various features such as number of days, average data surplus, associated data transfer object number and average data transfer amount carry out normalizing respectively
Change processing to obtain the data transfer object statistical nature after the corresponding normalized of each data transfer object, then will be each
It is special that data transfer object statistical nature after the corresponding normalized of data transfer object is built into data transfer object statistics
Levy matrix.
In the embodiment of the present application, it by obtaining the stream compression information of data transfer object, can be transferred the letter based on data flow
The stream compression managing detailed catalogue between at least two data transfer objects for including in breath constructs at least two data transfer objects
Between stream compression relationship, and extract at least two data transfer objects in each data transfer object data transfer object
Statistical nature.Then the corresponding neighbour of stream compression between at least two data transfer objects can be constructed according to stream compression relationship
Matrix is connect, it can structure further according to the data transfer object statistical nature of each data transfer object at least two data transfer objects
Build out data transfer object statistical nature matrix.Finally by the adjacency matrix constructed and data transfer object statistical nature matrix
Picture scroll product neural network model is inputted, can determine that at least two data turn based on the output result of picture scroll product neural network model
Move the data transfer object classification of each data transfer object in object.It, can be quick using method provided by the embodiments of the present application
Ground identifies data transfer object classification, improves the accuracy rate of data transfer object identification, strong applicability.
It is another process signal of data transfer object recognition methods provided by the embodiments of the present application referring to Figure 11, Figure 11
Figure.Data transfer object recognition methods provided by the embodiments of the present application can as follows 201 to 205 provide implementation
It is illustrated:
201, training sample is obtained, based on training sample training picture scroll product neural network model.
In some possible embodiments, the building of picture scroll product neural network model may include figure convolutional neural networks mould
The modeling data of type acquires, and picture scroll accumulates the data such as the training of neural network model and the test of picture scroll product neural network model
Processing stage.Wherein, the modeling data of the picture scroll product neural network model of acquisition can derive from local data base or cloud data
Library stores a large amount of training sample here in local data base or cloud database, for convenience of describing, can will acquire
Whole training samples are referred to as training sample set.It is understood that including multiple samples in each training sample obtained
The data transfer object classification of stream compression managing detailed catalogue and each sample data transfer object between data transfer object,
The data transfer object classification of sample data transfer object in middle training sample set may include two classes, and one kind is valid data
Transfer object, another kind of is invalid data transfer object, therefore available label 1 indicates valid data transfer object, and label 0 indicates
Invalid data transfer object, herein with no restrictions.It, can be according to each training for each training sample in training sample set
Stream compression managing detailed catalogue between the multiple sample data transfer objects for including in sample constructs multiple sample data transfers pair
The corresponding adjacency matrix of stream compression and data transfer object statistical nature matrix as between.
Specifically, the product of picture scroll defined in the embodiment of the present application neural network model includes input layer I, the first picture scroll product
Layer G1, the second picture scroll lamination G2, depth layer D3 and output layer O, and picture scroll product neural network model may particularly denote are as follows:
Here,Wherein, α is a hyper parameter, for balancing own node
The ratio of feature and neighbor node feature, D are the node degree matrix generated according to adjacency matrix A, and J is all sections in topological diagram
The unit matrix of point, W0∈RI×G1Indicate the weight matrix between input layer I to the first picture scroll lamination G1, W1∈RG1×G2Indicate figure
Weight matrix between convolutional layer G1 to picture scroll lamination G2, W2∈RG2×D3Indicate the weight square of picture scroll lamination G2 to depth layer D3
Battle array, W3∈RD3×OIndicate the weight matrix of depth layer D3 to output layer O.Here, the power in initial graph convolutional neural networks model
Value matrix W0~W3Dimension can be set by the user, and weight matrix W0~W3In the dimension of each weight matrix can be identical, or
Part is identical or entirely different, herein with no restrictions.It is understandable to be, based on the setting dimension got, if setting dimension
It is then that data transfer object statistical nature is mapped to height higher than the dimension of the data transfer object statistical nature matrix X of input
Dimension is to reflect data transfer object statistical nature if setting dimension is lower than the dimension of data transfer object statistical nature matrix X
It is mapped to low-dimensional.In addition, the weight matrix W in initial graph convolutional neural networks model0~W3Initial value can be by different initialization
Method determines.For example, if it is complete zero initial method, then weight matrix W0~W3Initial value be all 0, if it is Quan Yichu
Beginning method, then weight matrix W0~W3Initial value be all 1, if it is initial method is uniformly distributed, then weight matrix W0~
W3Value in the range of some setting, such as [- 0.5 ,+0.5], and meet and be uniformly distributed.Preferably, the application is implemented
Glorot initial method can be used in example, and this initial method is scholar Glorot and Bengio in Bengio AISTATS
A kind of initial method delivered on 2010, it is somewhat similar to be uniformly distributed initialization, but optimize the range of initial value value.
It specifically, can be by the sample data transfer object of building when carrying out the training of picture scroll product neural network model
The label of the data transfer object classification of adjacency matrix, data transfer object statistical nature matrix and label inputs initial graph convolution
The input layer I of neural network model, by input layer I to the mapping of the first convolutional layer G1, the first convolutional layer G1 to the second convolution
The mapping of layer G2 can be initial after the mapping of the second convolutional layer G2 to depth layer D3 and the mapping of depth layer D3 to output layer O
The prediction for the data transfer object classification that the output layer O of picture scroll product neural network model obtains any sample data transfer object is general
Rate, the error amount between the prediction probability and actual probabilities of any data transfer object based on output correct initial graph convolution
Model parameter (the weight matrix W of neural network model0~W3) with the corresponding adjoining of building input any data circulation managing detailed catalogue
Matrix and data transfer object statistical nature matrix can export the data transfer pair for including in any data circulation managing detailed catalogue
The picture scroll product neural network model of the data transfer object classification of elephant.After picture scroll product neural network model building is completed, it can adopt
Collect the adjacency matrix and data transfer object statistics of the test data transfer object of any several given data transfer object classifications
Test data of the eigenmatrix as picture scroll product neural network model.By the way that each test data is inputted the picture scroll that building is completed
Product neural network model, and will by picture scroll product neural network model output each test data transfer object prediction probability with
Actual probabilities are compared.If the data transfer object classification that picture scroll product neural network model is determined in multiple test datas
Number identical with real data transfer object classification is greater than or equal to preset times, illustrates the picture scroll product nerve net that building is completed
Network model meets building and requires, conversely, illustrating that the picture scroll product neural network model of building completion does not meet building and requires, then continues
The training of picture scroll product neural network model is carried out until meeting the requirements.Optionally, picture scroll product neural network model constructs completion
Afterwards, can also using loss function calculate training sample in all sample data transfer objects that label is marked cross entropy it
With if the sum of calculated cross entropy is less than or equal to default cross entropy, illustrate the figure convolutional neural networks mould that building is completed
Type meets building and requires, conversely, illustrating that the picture scroll product neural network model of building completion does not meet building and requires, then continues
The training of picture scroll product neural network model is until meeting the requirements.
Optionally, picture scroll product neural network model can also be trained using semi-supervised learning method.It is understandable to be, with
It is above-mentioned to be compared using supervised learning method, when using semi-supervised learning method training picture scroll product neural network model, the training sample of acquisition
Include in this set have label mark sample data transfer object data transfer object training sample and there is no label mark
The data transfer object training sample of the sample data transfer object of note, picture scroll accumulates the training process of neural network model not here
It is repeated again.
202, the stream compression information of data transfer object is obtained.
203, the stream compression relationship between at least two data transfer objects is constructed based on stream compression managing detailed catalogue, and
Extract the data transfer of each data transfer object in at least two data transfer objects for including in stream compression managing detailed catalogue
Object statistical nature.
204, the data transfer object classification of each data transfer object is identified based on picture scroll product neural network model.
Wherein, the specific implementation of step 202-204 may refer in the corresponding embodiment of Fig. 1 to step 101-103
Description, be not discussed here.
If 205, the data transfer object classification of any data transfer object is target data transfer object classification, control
The stream compression of the data transfer object.
In some possible embodiments, if the data transfer object classification of any data transfer object is target data
Transfer object classification can then be sent data transfer object stream compression abnormality alarming information by modes such as mail or phones and extremely be appointed
The data transfer object manager of one data transfer object is to remind data transfer object manager to the data transfer object
Stream compression is supervised.Alternatively, if the data transfer object classification of any data transfer object is target data transfer object
Classification then locks the stream compression of the data transfer object to block the data transfer object.In general, data transfer object
Classification includes two classes, and one kind is valid data transfer object, and another kind of is invalid data transfer object, and goal data turn
Moving object type is invalid data transfer object.By taking data transfer object is bank card as an example, when identifying certain bank card
It is the illegal bank card for money laundering, then the transmittable alarm email for carrying the bank card information to public security organ or bank card is sent out
To remind, public security organ personnel or bank clerk carry out risk control to the bank card to the public mailbox of row side or emphasis closes
Note, or can also directly freeze the fund flow of the bank card, it can specifically be determined according to practical application scene, not limited herein
System.
In the embodiment of the present application, by obtaining training sample, figure convolutional neural networks can be trained based on training sample
Model.It, can be based at least two numbers for including in stream compression information by obtaining the stream compression information of data transfer object
According to the stream compression managing detailed catalogue between transfer object, the stream compression relationship between at least two data transfer objects is constructed,
And extract the data transfer object statistical nature of each data transfer object at least two data transfer objects.According to data flow
The corresponding adjacency matrix of stream compression between at least two data transfer objects can be constructed by transferring the registration of Party membership, etc. from one unit to another, further according at least two numbers
It is special that data transfer object statistics can be constructed according to the data transfer object statistical nature of data transfer object each in transfer object
Levy matrix.Finally by the adjacency matrix constructed and data transfer object statistical nature Input matrix figure convolutional neural networks mould
Type can determine each data transfer at least two data transfer objects based on the output result of picture scroll product neural network model
The data transfer object classification of object, if the data transfer object classification of any data transfer object is target data transfer object
Classification can then control the stream compression of the data transfer object by alarm or other modes.It is provided using the embodiment of the present application
Method, can rapidly identify data transfer object classification, improve the accuracy rate of data transfer object identification, strong applicability.
It is the structural schematic diagram of data transfer object identification device provided by the embodiments of the present application referring to Figure 12, Figure 12.This
Application embodiment provide data transfer object identification device include:
Stream compression data obtaining module 31, for obtaining the stream compression information of data transfer object, above-mentioned data flow
It include the stream compression managing detailed catalogue between at least two data transfer objects in transfering the letter breath;
Data transfer object characteristic determination module 32, for what is determined based on above-mentioned stream compression data obtaining module 31
It states stream compression managing detailed catalogue and constructs stream compression relationship between above-mentioned at least two data transfer object, and extract above-mentioned number
According to the data transfer pair of each data transfer object in the above-mentioned at least two data transfer object for including in circulation managing detailed catalogue
As statistical nature;
Data transfer object identification module 33, for being based on picture scroll product neural network model, in conjunction with the transfer pair of above-mentioned data
Each data turn in the above-mentioned stream compression relationship and above-mentioned at least two data transfer object determined as characteristic determination module 32
The data transfer object statistical nature for moving object, determines each data transfer object in above-mentioned at least two data transfer object
Data transfer object classification.
In some possible embodiments, above-mentioned data transfer object characteristic determination module 32 includes:
Pipelined data acquiring unit 3201, for extracting the pipelined data in above-mentioned stream compression managing detailed catalogue;
Stream compression relationship construction unit 3202, for according to each data transfer object for including in above-mentioned pipelined data
Data transfer relationship constructs the stream compression relationship between above-mentioned at least two data transfer object;
Wherein, including at least data are transferred to, data produce, the data side of being transferred to and data connect in above-mentioned data transfer relationship
Debit.
In some possible embodiments, above-mentioned data transfer object identification module 33 includes:
Adjacency matrix construction unit 3301 is shifted for constructing above-mentioned at least two data according to above-mentioned stream compression relationship
The corresponding adjacency matrix of stream compression between object;
Data transfer object statistical nature matrix unit 3302, for according to every in above-mentioned at least two data transfer object
The data transfer object statistical nature of a data transfer object constructs data transfer object statistical nature matrix, wherein above-mentioned every
The data transfer object statistical nature of a data transfer object includes that stream compression number of days, stream compression enliven number of days, average
According at least one in surplus, associated data transfer object number and average data transfer amount;
Data transfer object classification determination unit 3303, for counting above-mentioned adjacency matrix and above-mentioned data transfer object
Eigenmatrix inputs picture scroll product neural network model, and the output result based on above-mentioned picture scroll product neural network model is determined
State the data transfer object classification of each data transfer object at least two data transfer objects.
In some possible embodiments, above-mentioned data transfer object statistical nature matrix unit 3302 is used for:
By the data transfer object statistical nature of data transfer object each in above-mentioned at least two data transfer object point
It is not normalized to obtain the corresponding normalization of each data transfer object in above-mentioned at least two data transfer object
Treated data transfer object statistical nature;
Based on the number after the corresponding normalized of data transfer object each in above-mentioned at least two data transfer object
Data transfer object statistical nature matrix is constructed according to transfer object statistical nature.
In some possible embodiments, above-mentioned data transfer object identification device further includes figure convolutional neural networks mould
Type determining module 34, wherein above-mentioned picture scroll accumulates neural network model determining module 34 and includes:
Training sample data acquiring unit 3401 includes multiple samples in above-mentioned training sample for obtaining training sample
The data transfer object classification of stream compression managing detailed catalogue and each sample data transfer object between data transfer object;
Training sample data processing unit 3402, for according to the data flow between above-mentioned multiple sample data transfer objects
Turn managing detailed catalogue and constructs the corresponding adjacency matrix of stream compression and data transfer pair between the multiple sample data transfer object
As statistical nature matrix;
Picture scroll accumulates neural network model training unit 3403, for the data based on above-mentioned multiple sample data transfer objects
Circulate the corresponding adjacency matrix of managing detailed catalogue, data transfer object statistical nature matrix and above-mentioned each sample data transfer object
Data transfer object classification, structure figures convolutional neural networks model so that above-mentioned picture scroll product neural network model have basis
Any data corresponding adjacency matrix of circulation managing detailed catalogue of input and data transfer object statistical nature Output matrix above-mentioned
The ability of the data transfer object classification for the data transfer object for including in one stream compression managing detailed catalogue.
In some possible embodiments, above-mentioned data transfer object identification device further include:
The first administration module of data transfer object 35, if turning for any data in above-mentioned at least two data transfer object
The data transfer object classification for moving object is target data transfer object classification, then it is abnormal to send data transfer object stream compression
Warning information reminds above-mentioned data transfer object management to the data transfer object manager of above-mentioned any data transfer object
Side supervises the stream compression of any of the above-described data transfer object.
In some possible embodiments, above-mentioned data transfer object identification device further include:
The second administration module of data transfer object 36, if turning for any data in above-mentioned at least two data transfer object
The data transfer object classification for moving object is target data transfer object classification, then locks any of the above-described data transfer object to hinder
The stream compression for any of the above-described data transfer object of breaking.
In the specific implementation, the device of above-mentioned data transfer object identification can be executed such as by each functional module built in it
Implementation provided by each step in above-mentioned Fig. 1 and Figure 11.For example, above-mentioned stream compression data obtaining module 31 can be used for
It executes and obtains stream compression information and other implementations in above-mentioned each step, for details, reference can be made to real provided by above-mentioned each step
Existing mode, details are not described herein.Above-mentioned data transfer object characteristic determination module 32 can be used for executing base in above-mentioned each step
The data of stream compression relationship and each data transfer object of extraction between stream compression information architecture data transfer object
Transfer object statistical nature and other implementations, for details, reference can be made to implementations provided by above-mentioned each step, no longer superfluous herein
It states.Above-mentioned data transfer object identification module 33, which can be used for executing, is based on picture scroll product neural network model in above-mentioned each step,
The data transfer object statistical nature of combined data circulation relationship and each data transfer object identifies each data transfer object
Data transfer object classification and other implementations, for details, reference can be made to implementations provided by above-mentioned each step, herein no longer
It repeats.Above-mentioned picture scroll product neural network model determining module 34, which can be used for executing in above-mentioned each step, obtains number of training
Accordingly and training picture scroll product neural network model and other implementations according to, processing number of training, for details, reference can be made to above-mentioned each steps
Implementation provided by rapid, details are not described herein.Above-mentioned the first administration module of data transfer object 35 can be used for executing above-mentioned
The stream compression for sending the warning information of invalid data transfer object in each step to supervise invalid data transfer object etc. is real
Existing mode, for details, reference can be made to implementations provided by above-mentioned each step, and details are not described herein.Above-mentioned data transfer object
Two administration modules 36 can be used for executing the stream compression and other implementations that invalid data transfer object is blocked in above-mentioned each step,
For details, reference can be made to implementations provided by above-mentioned each step, and details are not described herein.
In the embodiment of the present application, the device of data transfer object identification first can train figure by the training sample based on acquisition
Convolutional neural networks model can be based in stream compression information then by obtaining the stream compression information of data transfer object
Including at least two data transfer objects between stream compression managing detailed catalogue, construct at least two data transfer objects between
Stream compression relationship, and extract at least two data transfer objects in each data transfer object data transfer object statistics
Feature.The corresponding adjacency matrix of stream compression between at least two data transfer objects can be constructed according to stream compression relationship,
Number can be constructed further according to the data transfer object statistical nature of each data transfer object at least two data transfer objects
According to transfer object statistical nature matrix.Finally by the adjacency matrix constructed and data transfer object statistical nature Input matrix figure
Convolutional neural networks model can determine at least two data transfer objects based on the output result of picture scroll product neural network model
In each data transfer object data transfer object classification, if the data transfer object classification of any data transfer object be mesh
Data transfer object classification is marked, then can control the stream compression of the data transfer object by alarm or other modes.Using this
Apply for the method that embodiment provides, can rapidly identify data transfer object classification, improves the standard of data transfer object identification
True rate, flexibility is high, strong applicability.
It is the structural schematic diagram of terminal device provided by the embodiments of the present application referring to Figure 13, Figure 13.As shown in figure 13, this reality
Applying the terminal device in example may include: one or more processors 401 and memory 402.Above-mentioned processor 401 and memory
402 are connected by bus 403.For memory 402 for storing computer program, which includes program instruction, processing
Device 401 is used to execute the program instruction of the storage of memory 402, performs the following operations:
The stream compression information of data transfer object is obtained, includes that at least two data shift in above-mentioned stream compression information
Stream compression managing detailed catalogue between object;
The stream compression between above-mentioned at least two data transfer object is constructed based on above-mentioned stream compression managing detailed catalogue to close
System, and extract each data transfer pair in the above-mentioned at least two data transfer object for including in above-mentioned stream compression managing detailed catalogue
The data transfer object statistical nature of elephant;
Based on picture scroll product neural network model, in conjunction with above-mentioned stream compression relationship and above-mentioned at least two data transfer object
In each data transfer object data transfer object statistical nature, determine in above-mentioned at least two data transfer object each
The data transfer object classification of data transfer object.
In some possible embodiments, above-mentioned processor 401 is used for:
Extract the pipelined data in above-mentioned stream compression managing detailed catalogue;
According to the data transfer relationship for each data transfer object for including in above-mentioned pipelined data, building above-mentioned at least two
Stream compression relationship between data transfer object;
Wherein, including at least data are transferred to, data produce, the data side of being transferred to and data connect in above-mentioned data transfer relationship
Debit.
In some possible embodiments, above-mentioned processor 401 is used for:
The corresponding neighbour of stream compression between above-mentioned at least two data transfer object is constructed according to above-mentioned stream compression relationship
Connect matrix;
According to the data transfer object statistical nature of data transfer object each in above-mentioned at least two data transfer object
Construct data transfer object statistical nature matrix, wherein the data transfer object statistical nature of above-mentioned each data transfer object
Number of days, average data surplus, associated data transfer object number and average data are enlivened including stream compression number of days, stream compression
At least one of in transfer amount;
By above-mentioned adjacency matrix and above-mentioned data transfer object statistical nature Input matrix picture scroll product neural network model, and
Output result based on above-mentioned picture scroll product neural network model determines each data in above-mentioned at least two data transfer object
The data transfer object classification of transfer object.
In some possible embodiments, above-mentioned processor 401 is used for:
By the data transfer object statistical nature of data transfer object each in above-mentioned at least two data transfer object point
It is not normalized to obtain the corresponding normalization of each data transfer object in above-mentioned at least two data transfer object
Treated data transfer object statistical nature;
Based on the number after the corresponding normalized of data transfer object each in above-mentioned at least two data transfer object
Data transfer object statistical nature matrix is constructed according to transfer object statistical nature.
In some possible embodiments, above-mentioned processor 401 is used for:
Training sample is obtained, includes the stream compression detail between multiple sample data transfer objects in above-mentioned training sample
The data transfer object classification of information and each sample data transfer object;
The multiple sample number is constructed according to the stream compression managing detailed catalogue between above-mentioned multiple sample data transfer objects
According to the corresponding adjacency matrix of stream compression between transfer object and data transfer object statistical nature matrix;
The corresponding adjacency matrix of stream compression managing detailed catalogue, data transfer based on above-mentioned multiple sample data transfer objects
The data transfer object classification of object statistical nature matrix and above-mentioned each sample data transfer object, structure figures convolutional Neural net
Network model, so that above-mentioned picture scroll product neural network model has according to the corresponding adjoining of any data of input circulation managing detailed catalogue
The data transfer pair for including in matrix and any of the above-described stream compression managing detailed catalogue of data transfer object statistical nature Output matrix
The ability of the data transfer object classification of elephant.
In some possible embodiments, above-mentioned processor 401 is used for:
If the data transfer object classification of any data transfer object is target in above-mentioned at least two data transfer object
Data transfer object classification then sends data transfer object stream compression abnormality alarming information to above-mentioned any data transfer object
Data transfer object manager to remind above-mentioned data transfer object manager to the data of any of the above-described data transfer object
Circulation is supervised.
In some possible embodiments, above-mentioned processor 401 is used for:
If the data transfer object classification of any data transfer object is target in above-mentioned at least two data transfer object
Data transfer object classification locks any of the above-described data transfer object then to block the data flow of any of the above-described data transfer object
Turn.
It should be appreciated that in some possible embodiments, above-mentioned processor 401 can be central processing unit
(central processing unit, CPU), which can also be other general processors, digital signal processor
(digital signal processor, DSP), specific integrated circuit (application specific integrated
Circuit, ASIC), ready-made programmable gate array (field programmable gate array, FPGA) or other can
Programmed logic device, discrete gate or transistor logic, discrete hardware components etc..General processor can be microprocessor
Or the processor is also possible to any conventional processor etc..The memory 402 may include read-only memory and deposit at random
Access to memory, and instruction and data is provided to processor 401.The a part of of memory 402 can also include non-volatile random
Access memory.For example, memory 402 can be with the information of storage device type.
In the specific implementation, above-mentioned terminal device can execute such as above-mentioned Fig. 1 and Figure 11 by each functional module built in it
In implementation provided by each step, for details, reference can be made to implementations provided by above-mentioned each step, no longer superfluous herein
It states.
In the embodiment of the present application, terminal device first can train figure convolutional neural networks mould by the training sample based on acquisition
Type can be based on include in stream compression information at least two then by obtaining the stream compression information of data transfer object
Stream compression managing detailed catalogue between data transfer object, the stream compression constructed between at least two data transfer objects close
System, and extract the data transfer object statistical nature of each data transfer object at least two data transfer objects.According to number
The corresponding adjacency matrix of stream compression between at least two data transfer objects can be constructed according to circulation relationship, further according at least two
The data transfer object statistical nature of each data transfer object can construct data transfer object system in a data transfer object
Count eigenmatrix.Finally by the adjacency matrix constructed and data transfer object statistical nature Input matrix figure convolutional neural networks
Model can determine that each data turn at least two data transfer objects based on the output result of picture scroll product neural network model
The data transfer object classification of object is moved, if the data transfer object classification of any data transfer object is target data transfer pair
As classification, then the stream compression of the data transfer object can be controlled by alarm or other modes.It is mentioned using the embodiment of the present application
The method of confession can rapidly identify data transfer object classification, improve the accuracy rate of data transfer object identification, flexibility
Height, strong applicability.
The embodiment of the present application also provides a kind of computer readable storage medium, which has meter
Calculation machine program, the computer program include program instruction, which realizes each in Fig. 1 and Figure 11 when being executed by processor
Data transfer object provided by step knows method for distinguishing, for details, reference can be made to implementation provided by above-mentioned each step,
This is repeated no more.
Above-mentioned computer readable storage medium can be the dress for the data transfer object identification that aforementioned any embodiment provides
It sets or the internal storage unit of above-mentioned terminal device, such as the hard disk or memory of electronic equipment.The computer-readable storage medium
Matter is also possible to the plug-in type hard disk being equipped on the External memory equipment of the electronic equipment, such as the electronic equipment, intelligent storage
Block (smart media card, SMC), secure digital (secure digital, SD) card, flash card (flash card) etc..
Further, which can also both include the internal storage unit of the electronic equipment or deposit including outside
Store up equipment.The computer readable storage medium for other programs needed for storing the computer program and the electronic equipment and
Data.The computer readable storage medium can be also used for temporarily storing the data that has exported or will export.
Following claims and term " first " in specification and attached drawing, " second ", " third ", " the 4th " etc.
It is to be not use to describe a particular order for distinguishing different objects.In addition, term " includes " and " having " and they are any
Deformation, it is intended that cover and non-exclusive include.Such as contain the process, method, system, product of a series of steps or units
Or equipment is not limited to listed step or unit, but optionally further comprising the step of not listing or unit, or can
Selection of land further includes the other step or units intrinsic for these process, methods, product or equipment.
Referenced herein " embodiment " is it is meant that a particular feature, structure, or characteristic described can wrap in conjunction with the embodiments
It is contained at least one embodiment of the application.It is identical that each position in the description shows that the phrase might not be each meant
Embodiment, nor the independent or alternative embodiment with other embodiments mutual exclusion.Those skilled in the art explicitly and
Implicitly understand, embodiment described herein can be combined with other embodiments.In present specification and appended
Term "and/or" used in claims refers to any combination and institute of one or more of associated item listed
It is possible that combining, and including these combinations.Those of ordinary skill in the art may be aware that in conjunction with reality disclosed herein
Each exemplary unit and algorithm steps of example description are applied, can be come with the combination of electronic hardware, computer software or the two real
It is existing, in order to clearly illustrate the interchangeability of hardware and software, generally described in the above description according to function
Each exemplary composition and step.
Method provided by the embodiments of the present application and relevant apparatus be referring to method flow diagram provided by the embodiments of the present application and/
Or structural schematic diagram is come what is described, can specifically be realized by computer program instructions the every of method flow diagram and/or structural schematic diagram
The combination of process and/or box in one process and/or box and flowchart and/or the block diagram.These computer programs refer to
Enable the processor that can provide general purpose computer, special purpose computer, Embedded Processor or other programmable data processing devices
To generate a machine, so that generating use by the instruction that computer or the processor of other programmable data processing devices execute
In the function that realization is specified in one or more flows of the flowchart and/or structural schematic diagram one box or multiple boxes
Device.These computer program instructions, which may also be stored in, is able to guide computer or other programmable data processing devices with specific
In the computer-readable memory that mode works, so that it includes instruction that instruction stored in the computer readable memory, which generates,
The manufacture of device, the command device are realized in one box of one or more flows of the flowchart and/or structural schematic diagram
Or the function of being specified in multiple boxes.These computer program instructions can also be loaded into computer or the processing of other programmable datas
In equipment, so that executing series of operation steps on a computer or other programmable device to generate computer implemented place
Reason, so that instruction executed on a computer or other programmable device offer is for realizing in one process of flow chart or multiple
The step of function of being specified in process and/or structural representation one box or multiple boxes.
Claims (10)
1. a kind of data transfer object knows method for distinguishing, which is characterized in that the described method includes:
The stream compression information of data transfer object is obtained, includes at least two data transfer objects in the stream compression information
Between stream compression managing detailed catalogue;
Circulation managing detailed catalogue constructs the stream compression relationship between at least two data transfer object based on the data, and
Extract each data transfer object in at least two data transfer object for including in the stream compression managing detailed catalogue
Data transfer object statistical nature;
Based on picture scroll product neural network model, in conjunction with every in the stream compression relationship and at least two data transfer object
The data transfer object statistical nature of a data transfer object, determines each data in at least two data transfer object
The data transfer object classification of transfer object.
2. method according to claim 1, which is characterized in that the managing detailed catalogue building of circulation based on the data is described extremely
Stream compression relationship between few two data transfer objects, comprising:
Extract the pipelined data in the stream compression managing detailed catalogue;
According to the data transfer relationship for each data transfer object for including in the pipelined data, at least two data are constructed
Stream compression relationship between transfer object;
Wherein, include at least that data are transferred to, data produce, the data side of being transferred to and data receiver in the data transfer relationship
Side.
3. method according to claim 1 or claim 2, which is characterized in that it is described based on picture scroll product neural network model, in conjunction with described
The data transfer object of each data transfer object counts special in stream compression relationship and at least two data transfer object
Sign, determines the data transfer object classification of each data transfer object in at least two data transfer object, comprising:
The corresponding adjacent square of stream compression between at least two data transfer object is constructed according to the stream compression relationship
Battle array;
It is constructed according to the data transfer object statistical nature of data transfer object each in at least two data transfer object
Data transfer object statistical nature matrix, wherein the data transfer object statistical nature of each data transfer object includes
Stream compression number of days, stream compression enliven number of days, average data surplus, associated data transfer object number and average data transfer
At least one of in amount;
By the adjacency matrix and data transfer object statistical nature Input matrix picture scroll product neural network model, and it is based on
The output result of the picture scroll product neural network model determines each data transfer in at least two data transfer object
The data transfer object classification of object.
4. method according to claim 3, which is characterized in that described according to each in at least two data transfer object
The data transfer object statistical nature of data transfer object constructs data transfer object statistical nature matrix, comprising:
By the data transfer object statistical nature of data transfer object each in at least two data transfer object respectively into
Row normalized is to obtain the corresponding normalized of each data transfer object in at least two data transfer object
Data transfer object statistical nature afterwards;
Turned based on the data after the corresponding normalized of data transfer object each in at least two data transfer object
It moves object statistical nature and constructs data transfer object statistical nature matrix.
5. according to claim 3 or 4 the methods, which is characterized in that the method also includes:
Training sample is obtained, includes the stream compression managing detailed catalogue between multiple sample data transfer objects in the training sample
With the data transfer object classification of each sample data transfer object;
The multiple sample data is constructed according to the stream compression managing detailed catalogue between the multiple sample data transfer object to turn
Move the corresponding adjacency matrix of stream compression and data transfer object statistical nature matrix between object;
Based on the corresponding adjacency matrix of stream compression managing detailed catalogue of the multiple sample data transfer object, data transfer object
The data transfer object classification of statistical nature matrix and each sample data transfer object, structure figures convolutional neural networks mould
Type, so that picture scroll product neural network model has according to the corresponding adjacency matrix of any data of input circulation managing detailed catalogue
With the data transfer object for including in the circulation managing detailed catalogue of any data described in data transfer object statistical nature Output matrix
The ability of data transfer object classification.
6. any one of -5 the method according to claim 1, which is characterized in that the method also includes:
If the data transfer object classification of any data transfer object is target data in at least two data transfer object
Transfer object classification then sends the number of data transfer object stream compression abnormality alarming information to any data transfer object
According to transfer object manager to remind the data transfer object manager to the stream compression of any data transfer object
It is supervised.
7. any one of -5 the method according to claim 1, which is characterized in that the method also includes:
If the data transfer object classification of any data transfer object is target data in at least two data transfer object
Transfer object classification locks any data transfer object then to block the stream compression of any data transfer object.
8. a kind of device of data transfer object identification, which is characterized in that described device includes:
Stream compression data obtaining module, for obtaining the stream compression information of data transfer object, the stream compression information
In include at least two data transfer objects between stream compression managing detailed catalogue;
Data transfer object characteristic determination module, the data flow determined for the data obtaining module that circulates based on the data
Turn managing detailed catalogue and construct stream compression relationship between at least two data transfer object, and it is bright to extract the stream compression
The data transfer object statistics of each data transfer object is special in at least two data transfer object for including in thin information
Sign;
Data transfer object identification module, for being based on picture scroll product neural network model, in conjunction with the data transfer object feature
Each data transfer object in the stream compression relationship and at least two data transfer object that determining module determines
Data transfer object statistical nature determines that the data of each data transfer object in at least two data transfer object turn
Move object type.
9. a kind of terminal device, which is characterized in that including processor and memory, the processor and memory are connected with each other;
The memory is for storing computer program, and the computer program includes program instruction, and the processor is configured
For calling described program to instruct, the method according to claim 1 to 7 is executed.
10. a kind of computer readable storage medium, which is characterized in that the computer-readable recording medium storage has computer journey
Sequence, the computer program include program instruction, and described program instruction executes the processor such as
The described in any item methods of claim 1-7.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910323746.9A CN110209729A (en) | 2019-04-22 | 2019-04-22 | The method and device of data transfer object identification |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910323746.9A CN110209729A (en) | 2019-04-22 | 2019-04-22 | The method and device of data transfer object identification |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110209729A true CN110209729A (en) | 2019-09-06 |
Family
ID=67786194
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910323746.9A Pending CN110209729A (en) | 2019-04-22 | 2019-04-22 | The method and device of data transfer object identification |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110209729A (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111159481A (en) * | 2019-12-16 | 2020-05-15 | 深圳云天励飞技术有限公司 | Edge prediction method and device of graph data and terminal equipment |
CN111179085A (en) * | 2019-12-31 | 2020-05-19 | 中国银行股份有限公司 | Account data processing method, device and system |
CN113421066A (en) * | 2021-07-15 | 2021-09-21 | 郑州工程技术学院 | Method for monitoring school financial fund flow direction based on data tracking model |
CN113497729A (en) * | 2020-04-03 | 2021-10-12 | 中移动信息技术有限公司 | Training method of alarm recognition model, alarm recognition method and device |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104199832A (en) * | 2014-08-01 | 2014-12-10 | 西安理工大学 | Financial network unusual transaction community finding method based on information entropy |
CN108681553A (en) * | 2018-04-03 | 2018-10-19 | 平安科技(深圳)有限公司 | A kind of data processing method, device, terminal device and storage medium |
CN108960833A (en) * | 2018-08-10 | 2018-12-07 | 哈尔滨工业大学(威海) | A kind of abnormal transaction identification method based on isomery finance feature, equipment and storage medium |
CN109298926A (en) * | 2018-09-29 | 2019-02-01 | 阿里巴巴集团控股有限公司 | Resource transfers side enters the method, apparatus and electronic equipment of resource transfers platform |
-
2019
- 2019-04-22 CN CN201910323746.9A patent/CN110209729A/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104199832A (en) * | 2014-08-01 | 2014-12-10 | 西安理工大学 | Financial network unusual transaction community finding method based on information entropy |
CN108681553A (en) * | 2018-04-03 | 2018-10-19 | 平安科技(深圳)有限公司 | A kind of data processing method, device, terminal device and storage medium |
CN108960833A (en) * | 2018-08-10 | 2018-12-07 | 哈尔滨工业大学(威海) | A kind of abnormal transaction identification method based on isomery finance feature, equipment and storage medium |
CN109298926A (en) * | 2018-09-29 | 2019-02-01 | 阿里巴巴集团控股有限公司 | Resource transfers side enters the method, apparatus and electronic equipment of resource transfers platform |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111159481A (en) * | 2019-12-16 | 2020-05-15 | 深圳云天励飞技术有限公司 | Edge prediction method and device of graph data and terminal equipment |
CN111179085A (en) * | 2019-12-31 | 2020-05-19 | 中国银行股份有限公司 | Account data processing method, device and system |
CN111179085B (en) * | 2019-12-31 | 2023-06-20 | 中国银行股份有限公司 | Account data processing method, device and system |
CN113497729A (en) * | 2020-04-03 | 2021-10-12 | 中移动信息技术有限公司 | Training method of alarm recognition model, alarm recognition method and device |
CN113421066A (en) * | 2021-07-15 | 2021-09-21 | 郑州工程技术学院 | Method for monitoring school financial fund flow direction based on data tracking model |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110209729A (en) | The method and device of data transfer object identification | |
CN110717816A (en) | Artificial intelligence technology-based global financial risk knowledge graph construction method | |
CN110413707A (en) | The excavation of clique's relationship is cheated in internet and checks method and its system | |
CN110245787A (en) | A kind of target group's prediction technique, device and equipment | |
Zhan et al. | A loan application fraud detection method based on knowledge graph and neural network | |
CN110009475A (en) | Risk checks method for monitoring, device, computer equipment and storage medium | |
CN110414987A (en) | Recognition methods, device and the computer system of account aggregation | |
CN110443120A (en) | A kind of face identification method and equipment | |
CN109118053A (en) | It is a kind of steal card risk trade recognition methods and device | |
CN109727123A (en) | User's collage-credit data construction method, device and computer readable storage medium | |
CN109472626B (en) | Intelligent financial risk control method and system for mobile phone leasing service | |
CN111081337A (en) | Collaborative task prediction method and computer readable storage medium | |
CN110276618A (en) | The method and system for generating money laundering ancestor prediction model, predicting money laundering ancestor | |
CN107767055A (en) | A kind of mass-rent result assemblage method and device based on collusion detection | |
WO2022057108A1 (en) | Federated-learning-based personal qualification evaluation method, apparatus and system, and storage medium | |
CN108446988A (en) | User identification method based on relational network and system | |
Irarrázaval et al. | Telecom traffic pumping analytics via explainable data science | |
CN110188805B (en) | Identification method of fraud groups | |
CN114492827A (en) | Block chain technology-based federated learning model watermark reinforcement method and application | |
Chatterjee et al. | Next Generation Financial Services: Role of Blockchain enabled Federated Learning and Metaverse | |
CN115375480A (en) | Abnormal virtual coin wallet address detection method based on graph neural network | |
CN110059097A (en) | Data processing method and device | |
CN109447440A (en) | A kind of environmental objective appraisal management method and apparatus, terminal device and storage medium | |
CN109388747A (en) | The method and apparatus of the confidence level of user in a kind of acquisition network | |
CN110210884A (en) | Determine the method, apparatus, computer equipment and storage medium of user characteristic data |
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 |