CN106097095B - Determine the method and device of credit - Google Patents
Determine the method and device of credit Download PDFInfo
- Publication number
- CN106097095B CN106097095B CN201610409746.7A CN201610409746A CN106097095B CN 106097095 B CN106097095 B CN 106097095B CN 201610409746 A CN201610409746 A CN 201610409746A CN 106097095 B CN106097095 B CN 106097095B
- Authority
- CN
- China
- Prior art keywords
- geographical
- sequence
- geographical location
- place
- attribute
- 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.)
- Active
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
- G06Q40/02—Banking, e.g. interest calculation or account maintenance
Landscapes
- Business, Economics & Management (AREA)
- Accounting & Taxation (AREA)
- Finance (AREA)
- Engineering & Computer Science (AREA)
- Development Economics (AREA)
- Economics (AREA)
- Marketing (AREA)
- Strategic Management (AREA)
- Technology Law (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a kind of method and devices of determining credit;Method includes:Geographical location sequence is obtained based on data source, the geographical location sequence includes geographical location of the user residing for different time;It clusters the geographical location in the geographical location sequence to obtain place;The geographical attribute in each place is determined based on the corresponding geographical location in each place and time;Geographical attribute and corresponding time based on each place build geographical attribute sequence;The corresponding geographical attribute in each place in the geographical attribute sequence and time are carried out the first mapping to handle to obtain the credit feature of at least one dimension of the user;The credit feature of at least one dimension based on the user carries out the second mapping and handles to obtain the credit of the user.Implement the present invention, the information in the sequence of geographical location can be excavated to determine the credit of user comprehensively.
Description
Technical field
The present invention relates to field of computer technology more particularly to a kind of method and devices of determining credit.
Background technology
Credit investigation system is the credibility record established for user, is available to each bank, data subject, finance prison
Pipe mechanism, judicial department and other government organs use.The activity of this shared credit information is exactly reference.Credit investigation system one
Aspect is the tool prevented financial risk, and plays the role of safeguarding financial stability, on the other hand also plays propulsion social credibility
The effect of Establishing.
Currently, being based primarily upon collage-credit data (such as Bank Account Number flowing water, consumer record, credit for the credit for evaluating user
Record etc.), user the behavioral data of shopping platform with take in situation evaluate user credit.
But the relevant technologies evaluation user credit (can not evaluate the letter of all groups in the presence of low to the coverage rate of crowd
With) and collage-credit data it is not comprehensive and influence evaluation credit accuracy problems.
Invention content
The present invention provides a kind of method and device of determining credit at least to solve the above problem existing for the relevant technologies.
The technical proposal of the invention is realized in this way:
According to the first aspect of the invention, a kind of method of determining credit is provided, the method includes:
Geographical location sequence is obtained based on data source, the geographical location sequence includes ground of the user residing for different time
Manage position;
It clusters the geographical location in the geographical location sequence to obtain place;
The geographical attribute in each place is determined based on the corresponding geographical location in each place and time;
Geographical attribute and corresponding time based on each place build geographical attribute sequence;
The corresponding geographical attribute in each place in the geographical attribute sequence and time are carried out the first mapping to handle to obtain
The credit feature of at least one dimension of the user;
The credit feature of at least one dimension based on the user carries out the second mapping and handles to obtain the letter of the user
With.
According to an aspect of the present invention, a kind of device of determining credit is provided, described device includes:
Data acquisition module, for obtaining geographical location sequence based on data source, the geographical location sequence includes user
Geographical location residing for different time;
Geographical attribute constructing module, for clustering the geographical location in the geographical location sequence to obtain place;
The geographical attribute constructing module is additionally operable to be based on the corresponding geographical location in each place and the time is true
The geographical attribute in fixed each place;Geographical attribute and corresponding time based on each place build geographical attribute sequence;
The geographical attribute constructing module, be additionally operable to by the corresponding geographical attribute in each place in the geographical attribute sequence with
And the time carries out the first mapping and handles to obtain the credit feature of at least one dimension of the user;
Credit mapping block, the credit feature at least one dimension based on the user carry out the second mapping processing
Obtain the credit of the user.
The invention has the advantages that:The attribute of different location, base are determined using the data in the geographical location of user
Attribute in different location maps the credit feature of (the first mapping processing) user, and then maps (the second mapping processing) user's
Credit, on the one hand realize to the multi-angle of user's life time cover to further increase credit scoring accuracy and can
By property, the boundary for the crowd's covering that can evaluate credit has on the other hand also further been widened.
Description of the drawings
Fig. 1 is an optional hardware architecture diagram of the device that credit is determined in the embodiment of the present invention;
Fig. 2-1 is the optional application scenarios schematic diagram that information is determined in the embodiment of the present invention;
Fig. 2-2 is the optional application scenarios schematic diagram that information is determined in the embodiment of the present invention;
Fig. 3 is an optional flow diagram of the method that credit is determined in the embodiment of the present invention;
Fig. 4 is another the optional flow diagram for the method that credit is determined in the embodiment of the present invention;
Fig. 5 is another the optional flow diagram for the method that credit is determined in the embodiment of the present invention;
Fig. 6 is another the optional flow diagram for the method that credit is determined in the embodiment of the present invention;
Fig. 7, which is that one of matrix elasticity partition-merge in the embodiment of the present invention is optional, realizes schematic diagram;
Fig. 8 is that multidimensional data matrix is split as location data module and multidimensional location data module in the embodiment of the present invention
Schematic diagram;
Fig. 9 is the optional flow diagram that static geographical attribute is extracted in the embodiment of the present invention;
Figure 10 is the optional flow diagram that Dynamic Geographic attribute is extracted in the embodiment of the present invention;
Figure 11 is an optional structural schematic diagram of the device that credit is determined in the embodiment of the present invention.
Specific implementation mode
First to the relevant technologies evaluate user credit mode there are the problem of illustrate, inventor implement the present invention
During find, in the related technology evaluate user credit have at least the following problems:
1) crowd of collage-credit data covering is limited
In traditional reference mode as an example, it is based primarily upon finance activities on the lines such as bank's flowing water and the passing credit record of user
Data evaluation credit, the data class that uses is less, data volume is small and low frequency, and collage-credit data is low to the coverage rate of crowd,
Since current a considerable amount of crowds do not have collage-credit data, there is no the crowd of collage-credit data that can not evaluate credit these.
2) accuracy of credit evaluation is limited
By taking behavioral data Internet-based evaluates credit as an example, the behavioral data based on user in internet evaluates letter
With, such as it is right in the behavioral data (such as do shopping, collect, the data of payment, goods browse behavior) of shopping platform based on user
The carry out credit evaluation of user.
Again by taking the income situation evaluation credit based on user as an example, based on personal profession income, industry average income, region
Average income etc. counts class data to assess the credit grade of certain user, collects and is explicitly come pair using income/expenditure data
The repaying ability of user is assessed, and then calculates credit risk.
In summary analysis can find out that the relevant technologies are in bank's flowing water, credit record and shopping etc. and money with volume
The aspect of correlated activation collects collage-credit data, the defect that there are types is single for collected collage-credit data, data volume is few, due to only
Can certain customers collect collage-credit data (such as the collage-credit data of the above-mentioned type pair can not be collected with substantial portion of rural subscribers),
Cause the coverage rate of credit evaluation low, and exists on the credit evaluation of user and because collage-credit data is not comprehensive influence accuracy
Problem.
The collage-credit data that the relevant technologies are collected be to user be engaged in shopping and finance etc. it is directly related with money it is movable into
Row data collection obtains, and these activities only occupy the ratio of very little in the life cycle of user, therefore only to user
The moving collection collage-credit data of the very little ratio of life cycle.
Inventor has found in the practice of the invention, for equal and golden with shopping and finance in the life cycle of user
The activity or information that money is not directly linked, geographical location (such as residence, place of working, public place of entertainment), outdoor fortune such as user
The relevant feature of credit with user such as loan repayment capacity and refund wish of user can be reflected indirectly by moving, and the geography of user
Position and corresponding time (time when being in the geographical location) are the major parts that can be directed to whole mobile terminal users
Life cycle is collected easily.
For being based on mobile terminal and collect geographical location, the popularity rate of mobile terminal nearly reaches 100%, beyond correlation
The ratio that the user of collage-credit data is collected in technology, it is (and corresponding to the geographical location of some region of mobile terminal user
Time) be collected and can be considered that the geographical location to whole crowds of this area is collected, mobile terminal user uses base
User's progress is also had exceeded in the time of the application such as social networking application of location-based service (LBS, Location Based Service)
Finance, consumption movable time, by obtaining geographical location and time of the mobile terminal user using application, compared with the relevant technologies
Realization to subscriber lifecycle more movable acquisition, based on these movable geographic position datas (including geographical location with
And the corresponding time) can obtain more comprehensively and the relevant feature of the credit of user.
To which once obtaining the position data of user, then most of life cycle that can be based on user (is given birth to beyond user
Finance, consumption movable time are engaged in the life period, because only having a small amount of time in the life cycle of user for doing shopping
And finance activities) in the static geographical attribute in residing geographical location, the Dynamic Geographic attribute that is moved between diverse geographic location
User credit is evaluated, the coverage rate for evaluating the user of credit is wider (to acquire movement compared with the coverage rate of the relevant technologies reference user
For terminal user is using the geographical location of wechat, for substantial portion of wechat user such as rural resident and the middle-aged and the old, only
Collect position data and collage-credit data can not be collected), therefore, the geographical position based on most of life cycle acquisition in user
The credit that data are evaluated is set, the credit compared with the collage-credit data evaluation of the fraction life cycle acquisition in user can more comprehensively
Accurately.
Before the present invention will be described in further detail, to involved in the embodiment of the present invention noun and term say
Bright, noun and term involved in the embodiment of the present invention are suitable for following explanation.
1) finance activities data (such as Bank Account Number of collage-credit data, the data of the credit for evaluating user, including user
Flowing water, credit record), the behavioral data of shopping platform (such as do shopping, collect, the data of payment, goods browse behavior) and receipts
Enter situation (as the personal income of user, user are engaged in region average income residing for industry average income, user).
2) credit can be continuous score value (credit scoring), can also be discrete credit grade, can be used for weighing and use
The credit risk at family and the ability repaid the loan is fulfiled, the credit of user is higher, then the ability repaid the loan is higher, promise breaking
Probability is lower.
3) geographical location, user may be used longitude and latitude etc. and arbitrarily may be used the location of when using based on location-based service
It is characterized in a manner of demarcating position.
4) place obtains after being clustered to geographical location, and place can be a geographical location, can also be a region.
5) geographical location sequence is the sequence that basic element is constituted with " when m- geographical location ", and geographical location refers to user
The location of when using based on location-based service, the time refer to user be in the position time (can be a certain moment, also may be used
Think that one is the period).Illustratively, using form recording geographical position as " geographical location marker-time " and
Time.
5) geographical attribute, including:
Static geographical attribute (static geographical attribute):The classification category corresponding to several places that user often haunts
Property.
Dynamic geographical attribute (Dynamic Geographic attribute):Motion track pattern (such as place of the user between multiple places
The places A- B- place C) or space-time migration model (for example, " 6 points of every morning adheres to outdoor sports ", " every night 8 points or so come off duty
Driving is gone home " etc.);In the classification system of the different mode pre-established, to the same pattern (motion track pattern and space-time
Migration model) it carries out the division (description) of multiple dimensions and assigns respective labels, obtain motion track pattern or space-time migration mould
The multidimensional of formula describes.
6) (App) is applied:The application software being often referred in the narrow sense in equipment (such as smart mobile phone), also refers to all calculating
All application software and its son on machine equipment (containing PC, mobile terminal, cloud computing sever platforms etc.) except division operation system are soft
Part (such as plug-in unit).
The present invention is further described in detail below with reference to the accompanying drawings and embodiments.It should be appreciated that mentioned herein
Embodiment is only used to explain the present invention, is not intended to limit the present invention.In addition, embodiment provided below is for implementing
The section Example of the present invention, rather than the whole embodiments for implementing the present invention are provided, in the absence of conflict, the present invention is implemented
Example record technical solution can be in any combination mode implement.
It should be noted that the term " first second third " involved by the embodiment of the present invention is only to be that difference is similar
Object, do not represent the particular sorted for object, it is possible to understand that ground, " Yi Er thirds " can be in the case of permission
Exchange specific sequence or precedence.It should be appreciated that the object that " first second third " is distinguished in the appropriate case can be mutual
It changes, so that the embodiment of the present invention described herein can be real with the sequence other than those of illustrating or describing herein
It applies.
The embodiment of the present invention can be provided as determining the method for credit and determine the device of credit, in practical application, determine
Each function module in the device of credit can by the hardware resource of equipment (such as terminal device, server or server cluster),
Such as processor computing resource, the communication resource (being such as used to support to realize that optical cable, the various modes of honeycomb to communicate) cooperative achievement.Figure
1 illustrates an optional hardware architecture diagram of equipment 10, including processor 11,13 (example of input/output interface
Such as one or more of display, keyboard, touch screen, Speaker Microphone), storage medium 14 and network interface 12, group
Part can be through 15 connection communication of system bus.
Certainly, the embodiment of the present invention is not limited to providing method and hardware.For example, as shown in Fig. 2-1, in practical application,
The embodiment of the present invention can be provided as execute credit determine method software function module (including it is a series of for execute credit determination
The executable instruction of method), to be coupled to existing arbitrary application such as social networking application, credit application, answered in use for user
The querying individual credit during, certainly, in addition to being realized in a manner of software function module, as shown in Fig. 2-2, the present invention is real
It applies example also and can be provided separately as credit evaluation platform, in a specific way such as application programming interfaces (API, Application
Interface), plug-in unit or software development kit (SDK, Soft Development Toolkit) mode provide public credit
The calling for inquiring service, so that enterprise, entity and individual make a credit inquiry.
It should be pointed out that the software function module that the embodiment of the present invention is provided can be embedded into comprehensive social networks number
According to reference points-scoring system in, with combined with other credit scoring strategies in reference points-scoring system determine user credit, a side
Face, the credit scoring model based on position data are the important supplements of all multi-models based on other social network datas, can be with
Significantly improve the accuracy and reliability of reference points-scoring system output result;On the other hand, for Mr. Yu class user, if should
The social network data of class user is excessively sparse, does not cover, at this time can be with if geographic position data has certain abundance
Rely primarily on the such user credit evaluation of output of the credit scoring model based on position data.Certainly, it is based on geographical location number
According to credit scoring model be used alone when, it may have excellent credit scoring ability can take out for credit operation
Candidate crowd of the higher head user of credit scoring as credit is taken, the crowd for expanding general favour finance covers boundary.
The embodiment of the present invention determines that the credit of user may include three phases in the following ways:
The first, geographical location sequence is formed.
The second, intermediate result namely geographical attribute for determining user credit are obtained based on geographical location sequence, including
Static geographical attribute and Dynamic Geographic attribute.
Third carries out mapping processing based on intermediate result (illustratively, subsequently to carry out successively in the embodiment of the present invention
Illustrated for one mapping processing and the second mapping processing processing of mapping twice) obtain the credit of user.
Below in conjunction with the method for the determination credit shown in Fig. 3 an optional flow diagram to the above three stage
Processing illustrates, and determines that the method for the credit of user at least includes the following steps in figure 3.
Step 101, it is based on data source and obtains geographical location sequence.
Geographical location sequence includes geographical location of the user residing for different time.
Step 102, it clusters the geographical location in the sequence of geographical location to obtain place.
Step 103, the geographical attribute in each place is determined based on the corresponding geographical location in each place and time.
Step 104, geographical attribute and corresponding time based on each place build geographical attribute sequence.
Step 105, the corresponding geographical attribute in each place in geographical attribute sequence and time are subjected to the first mapping processing
Obtain the credit feature of at least one dimension of user.
Step 106, credit feature the second mapping of progress of at least one dimension based on user handles to obtain the letter of user
With.
Illustratively, a plurality of types of equipment may be used in data source, such as:Mobile phone, tablet computer, Wearable (intelligence
Energy wrist-watch, intelligent glasses) and car-mounted terminal etc., above equipment has locating module, and is provided with the permission to terminal positioning
So as to collect geographical location and the time of relative users.Alternatively, being based on location-based service (such as social networking application in terminal operating
The navigation feature etc. searched in nearby friends functions, map application).
In some embodiments, in a step 101, the geographical location sequence of user is obtained only from a data source, and
The geographical attribute for parsing each geographical location in the sequence of geographical location in step 103, to build geographical attribute at step 104
Intermediate result of the sequence as user's map user credit.Such as the location data of the smart mobile phone from user obtains geographical location
Sequence, and parse each geographical location in geographical position sequence and obtain corresponding geographical attribute.
In some embodiments, in a step 101, it is the credit for more accurately determining user comprehensively, from multiple data sources
(such as smart mobile phone, tablet computer and car-mounted terminal) obtains geographical location sequence one by one, then is formed for waiting solving in a step 102
The geographical location sequence of analysis, and form geographical attribute sequence in step 102 to step 105 and be mapped as having when the credit of user
A variety of processing modes below illustrate different processing modes.
Mode one) it corresponds to form multiple intermediate results from the geographical location sequence of multiple data sources acquisition, fusion is more
A intermediate result map user credit.
For corresponding to the multiple geographical location sequences obtained from multiple data sources, directly each geographical location sequence is distinguished
It is parsed to obtain the geographical attribute in each geographical location, forms geographical attribute sequence corresponding with each geography position sequence, as
The intermediate result of map user credit, using the intermediate result of each data source of correspondence carry out mapping processing (first mapping processing and
Second mapping is handled) obtain user credit.
One example is as shown in figure 4, be respectively processed the collected geographical location sequence of each data source to obtain phase
The geographical attribute sequence answered:Geographical location sequence of the parsing from data source 1 obtain intermediate result 1 (namely geographical attribute sequence,
Such as static geographical sequence of attributes, Dynamic Geographic sequence of attributes form), an intermediate result as map user credit;Together
Reason, geographical location sequence pair of the parsing from data source 2, data source 3 should obtain intermediate result 2 and intermediate result 3, as mapping
The intermediate result of user credit.
Continue to illustrate Fig. 4, after the intermediate result for obtaining corresponding to each data source, by intermediate result (intermediate result
1, intermediate result 2 and intermediate result 3), that is, the corresponding geographical attribute in each place and time utilization in geographical attribute sequence
Mapping model carries out mapping processing (carrying out the first mapping processing and the second mapping processing successively), and the credit for obtaining user credit is commented
Divide result (it is of course also possible to use the forms such as grading of credit).
In some embodiments, when be combined the intermediate result that multiple data sources obtain carry out mapping processing when, consider
To the reliability, accuracy and sampling density of the data of different data sources (data source acquires frequency to the data in geographical location)
Difference, in order to ensure the confidence level of determining user credit, to the centre from different data sources in credit mapping model
As a result weight is distributed, that is, maps the geographical sequence of attributes distribution weights of the difference handled for pending first and (characterizes for reflecting
Penetrate the significance level of user credit), the intermediate result map user credit of the corresponding multiple data sources of fusion, credit mapping model can
The intermediate knot of corresponding multiple data sources is merged in a manner of using linear weighted function, complex nonlinear weighting (such as neural network) etc.
Fruit.The corresponding weight of geographical attribute sequence is (namely determines that geographical sequence of attributes is used geographical location based on derived data source
Derived data source) reliability and at least one accuracy determine.
Still by taking Fig. 4 as an example, if the reliability of data source 1, data source 2 and data source 3 reduces successively, to for intermediate result
1, intermediate result 2 and intermediate result 3 correspond to distribution weight 1, weight 2 and weight 3, wherein weight 1>Weight 2>Weight 3 so that in
Between result 1, intermediate result 2 and intermediate result 3 influence degree of user credit is sequentially reduced, it is ensured that determine user credit can
Reliability.
Mode two) it is overlapped from the geographical location sequence of multiple data sources acquisition, utilize the geographical location sequence after superposition
Row form intermediate result, utilize intermediate result map user credit.
Above-mentioned multiple data sources can be different types of multiple data sources, and multiple data sources here can be from multiple
Dimension is divided.For example, being divided from the dimension of hardware device, different hardware devices is divided into different types of number
According to source.In another example being divided from the dimension of software, different software such as App corresponds to different types of data source.For another example
The dimension of different data collection point (operation for the triggering geographic position data acquisition that user implements) divides from same application
For different types of data source.
For corresponding to the multiple geographical location sequences obtained from multiple data sources, by the same time in the sequence of geographical location
Geographical location be overlapped, in the geographical location sequence after superposition geographical location cluster be place, obtain the ground in each place
Attribute is managed, geographical attribute (static geographical attribute and Dynamic Geographic attribute) and time based on each place build geographical attribute
Sequence carries out mapping processing using intermediate result as the intermediate result for map user credit in credit mapping model
Obtain user credit.
In some embodiments, in order to avoid coming from the geographical position in corrupt data source (or accuracy not high data source)
Set the negative effect that sequence pair determines the accuracy of user credit, multiple geographical location sequences are overlapped to be formed it is to be resolved
It is corresponding different in the integrated data sequence of geographical location when geographical location sequence (also referred to as geographical location integrated data sequence)
Weight is distributed in the geographical location of data source, weight be according to the reliability, accuracy and sampling density of each data source data at least
One of determine so that based on the appraisal result for the user credit that the geographical location sequence pair of more reliable data source output determines
Influence bigger, ensure the confidence level of user credit mapped out.
An example as superposition geographical location sequence, it is assumed that have 10 data sources that can collect the geographical position of user
It sets, in the geographical location for thering are 3 data sources (such as smart mobile phone, car-mounted terminal and smartwatch) to collect user at 7 o'clock, that
After being overlapped for the geographical location sequence of 3 data sources output, in 3 geographical locations for being corresponding with user 7 o'clock
The packing density of data, the geographical location of same time increases.
Another example obtains as shown in figure 5, by the collected geographical location sequence of each data source based on sequential superposition
Geographical location integrated data sequence, geographical location integrated data sequence each time include that multiple data sources are collected in the time
User geographical location, thus the packing density in the geographical location in each time be more than data mapping.Based on geographical position
The geographical attribute that integrated data sequence analysis goes out each place is set, such as static geographical attribute and Dynamic Geographic attribute, if static geographical
Attribute and Dynamic Geographic attribute occupy the classification of multiple dimensions, then form the quiet Dynamic Geographic sequence of attributes of multidimensional and multidimensional statically
Sequence of attributes is managed as intermediate result, credit is carried out in credit mapping model based on intermediate result and maps to obtain user credit,
Illustratively, it obtains final credit using the classification of linear and complex nonlinear and regression algorithm in credit mapping model and comments
Point.
Mode three) by the geographical location sequence construct multidimensional data matrix (HDLM) from multiple data sources, from multidimensional number
According to matrix extraction location data block (LDB, Location Data Block) and multidimensional location data block (MDB, Multiple
Dimensional Location Data Block), to location data block and multidimensional location data geographical location sequence in the block
It carries out dissection process and obtains geographical attribute sequence (the static geographical sequence of attributes of such as multidimensional, Dynamic and Multi dimensional geographical attribute sequence shape
Formula) as the intermediate result for map user credit, mapping is carried out to intermediate result using credit mapping model and handles to obtain
User credit.
One example is as shown in fig. 7, the geographical location sequence that multiple data sources are exported, according to each geographical location (with reality
Line box identifies) sequential (namely each geographical location corresponds to the sequencing of time) alignment, if data source exports
Geographical location sequence in corresponding geographical location sometime lack (this is because data source does not collect use in the time
The geographical location at family), then the value being calculated using the value (such as zero, identified with dashed rectangle) or interpolation algorithm of acquiescence is filled
For the geographical location of missing, will alignment and filling treated each geographical position sequence as row vector structure multidimensional data square
Battle array.
Every a line of multidimensional data matrix shown in Fig. 8 corresponds to the geographical location sequence from a data source and (passes through
The alignment of sequential, the filling for lacking geographical location and interpolation processing), the geographical location sequence of multiple data sources is carried out first
The processing of matrix elasticity partition-merge obtains the row vector (also referred to as primary features sequence) of structure multidimensional data matrix, then base
Geographical attribute (including the static geographical attribute and dynamic of each place (clustering to obtain to geographical location) is extracted in multidimensional data matrix
State geographical attribute) to build geographical attribute sequence (also can be considered secondary features sequence), as in map user credit
Between result.Finally geographical attribute sequence is handled in credit mapping model by the classification of linear and complex nonlinear and recurrence
(as used regression algorithm), obtains the credit scoring result of user.
The place of primary features sequence is obtained into row matrix elasticity partition-merge to multiple geographical location sequences to above-mentioned again
Reason illustrates.
Usually, only have a data source that can collect geographical location sometime in multiple data sources, accordingly
Ground, in multidimensional data matrix each row (vector) usually only certain one-dimensional geographical location there are Effective Numerical, the column vector
(can fill default value) that the data in the geographical location of other dimensions are missing from has a large amount of only one in multidimensional data matrix
A dimension has a column vector of value, and such column vector have the characteristics that it is continuously distributed, to form several big blocks
It is exactly location data block.
In addition, in multidimensional data matrix other than there are big block, there is also block of cells i.e. multidimensional location datas
Block, block of cells are made of the continuous column vector that all there is the Effective Numerical in geographical location in multiple dimensions.For example, setting presence
10 data sources collect the position of user at continuous three moment simultaneously, and (each dimension corresponds to one to 10 dimensions at the moment
A data source) geographical location constitute multidimensional data matrix in a column vector, each column vector includes the ground of 10 dimensions
Position is managed, continuous three time, corresponding column vector formed a multidimensional location data block in multidimensional data matrix.
In some embodiments, as shown in figure 8, identifying the location data in multidimensional data matrix by following mode
Block and multidimensional location data block:Identification multidimensional data matrix column vector is scanned by column, identifying in poly-dimensional block data only has
The continuous column vector of effective geographic position data of one dimension is location data block;Identifying in poly-dimensional block data has at least
The continuous column vector of effective geographic position data of two dimensions is multidimensional location data block.Entire multidimensional data matrix is divided
Sequence (primary features sequence) for the location data block and multidimensional location data block alternative splicing that staggeredly splice, wherein positioning number
Inconsistent (the acquisition in each column vector corresponding data source of quantity according to the column vector included by block, multidimensional location data block
Time), therefore the length of time that location data block and multidimensional location data block are covered is inconsistent.
In some embodiments, for the location data block identified from multidimensional data matrix, by each location data block
In geographical location sequence based on sequential be superimposed, using the ground that the geographical location sequence obtained after superposition is to be resolved as step 103
Position sequence is managed, the geographical attribute in each place is gone out based on geographical location integrated data sequence analysis, forms quiet Dynamic Geographic attribute
Sequence and static geographical sequence of attributes carry out credit in credit mapping model based on intermediate result and map as intermediate result
To user credit.
In addition, in order to ensure the confidence level of determining user credit, weight can be distributed for different location data blocks,
At least one weight and the reliability in location data block corresponding data source and accuracy are determining, illustratively, location data block
Reliability and accuracy positive correlation of the weight with corresponding data source, the reliability and accuracy of data source are higher, then corresponding positioning
The weight of data block is higher, so that the user determined based on the geographical location sequence pair of more reliable data source output is believed
The influence bigger of appraisal result, ensures the confidence level of user credit mapped out.
In some embodiments, for the multidimensional location data block identified from multidimensional data matrix, since multidimensional positions
Data block is in the same time while to be collected the geographical location of user due to multiple data sources and formed, and not only the time is corresponding
The packing density in geographical location is higher than the packing density in geographical location in location data block, but also illustrates that user is now in certain
A little special scenes implement certain specific behaviors.
For example, the vehicle mounted guidance App of user collects geographical location, while the mobile phone of user is also adopted in masses' comment App
Collect geographical location, illustrates that user is possible to driving and searching where to have a meal.If some good friend of the user at this time
Mobile phone also collect geographic position data, and the geographical location of good friend and the position of this user are very synchronous, then are likely to
It is that user drives to carry good friend and plans to go somewhere to have a meal together.
Correspondingly, when determining in step 103 based on multidimensional location data block construction geographical attribute sequence, multidimensional is positioned
Geographical location sequence in data is mapped to semantic classes system (including multiple dimensions of scene or behavior residing for characterization user
Classification) in, obtain each geographical location multiple dimensions Dynamic Geographic attribute.
Again to forming corresponding geographical attribute sequence simultaneously based on geographical location sequence in three kinds of mode either types above-mentioned
The processing of map user credit illustrates, and the geographical attribute in geographical location includes static geographical attribute and Dynamic Geographic attribute,
Correspondingly, geographical attribute sequence includes static geographical sequence of attributes and Dynamic Geographic sequence of attributes, separately below to combining static state
Geographical attribute map user credit and combination Dynamic Geographic attribute map user credit explanation.
1) static geographical attribute
In one embodiment, the geographical location in the sequence of geographical location is clustered (such as poly- based on distribution density
Class or cluster based on Euclidean distance) it is place, place can be a geographical location, can also be by multiple geographical location structures
At region (such as the region formed by multiple geographical locations).
The label that corresponding geographical location in geographical position sequence is replaced using the label in the place after cluster, when obtaining place
Between sequence (binary combination of place and time constitute sequence), such as geographical location sequence:The geographical location 1- times 1,
Position 2- times 2, geographical location 3- times 3, geographical location 4- times 4 are managed, if geographical location 1 and the cluster of geographical location 2 are ground
Point 1, geographical location 3 and the cluster of geographical location 4 are place 2, then use place 1 to replace geographical location 1 and geographical location 2, utilize ground
2 replacement geographical location 3 of point and geographical location 4, the location sequence of formation are:When place 1- times 1, place 1- times 2, place 2-
Between 3, the place 2- times 4.
After forming place time series, in the time series of place each place and the corresponding time from least one
A dimension carries out semantic classes classification, obtains the static geographical attribute that each place in the time series of place corresponds at least one dimension
(multidimensional static state geographical attribute is then formed when multiple dimensions), based on the static geographical attribute in each place, the corresponding time shape in place
At static geographical sequence of attributes, the form as static geographical attribute 1- times 1, static geographical attribute 2- times 2.
An example of static geographical sequence of attributes is formed as shown in figure 9, geographical location for random length (being denoted as n)
Sequence passes through density-based algorithms (such as DBSCAN etc.) or the clustering algorithm based on Euclidean distance (such as K-means
Deng), it obtains several and (is denoted as m, lists of the usual m much smaller than n) place (or region).Place is tieed up based on m, and geographical position is tieed up to n
The label for setting geographical location in sequence is replaced, and obtains n dimension place time serieses.
The place in the time series of place is tieed up to n by the way of semantic classes classification and carries out semantics recognition, is obtained static
Geographical attribute sequence:M ties up " place-classification " list.Every a line in m dimension " place-classification " lists has 2 data item, and first
A data item is place ID, and second data item is that the obtained classification of semantics recognition, such as residence, work are passed through in place
Ground, amusement and leisure etc. classifications.
It should be pointed out that when carrying out semantics recognition classification to the various regions point in the time series of place, one can be used
(semantics recognition grader is for each place in the time series of place for semantic classes grader or multiple semantics recognition graders
Carry out the mathematical model of semantics recognition classification).To each using multiple semantic classes graders (namely from multiple dimensions)
When point carries out semantics recognition classification, correspondingly, m dimension " place-classification " lists have also just been extended to multiple row vector namely matrix
Form, each column vector correspond to possessed classification of each place under some taxonomic hierarchies, an example such as 1 institute of table
Show:
Table 1
As it can be seen from table 1 same place label with different classifications under different taxonomic hierarchies, and it is certain
The label of classification under dimension can lack (for example, because data source does not collect corresponding geographical location or because class
Type is excessively fuzzy and can not definitely classify).
2) credit mapping is handled
The processing to each place of determination in the static geographical attribute of at least one dimension is connected to illustrate.Geographical attribute sequence
The corresponding geographical attribute in each place and time carry out the first mapping processing in row, illustratively, statically based on each place
Reason attribute and user are in the time in all kinds of places, by the Nonlinear Classifier and regression model built in advance, determine at least
One dimension, the direct or indirect credit feature of reflection user's loan repayment capacity and refund wish, illustratively, including:User
Income level, the level of consumption, job specification (such as day shift, night shift or in shifts), occupation type (such as high-tech enterprise,
School, institutional settings etc.), quality of life and life health degree be (for example whether often staying up late for work, living or working for a long time
Pressure is big etc.).
The second mapping processing is carried out to credit feature based on the feature that mapping obtains, then by compressive classification and regression model
Obtain final credit scoring result.
The processing of the above-mentioned map user credit based on static geographical attribute is provided that be credit mapping block, in addition to can
To be used alone and outside export credit appraisal result, can also will reflect that multiple dimensions of user's loan repayment capacity and refund wish are special
Sign is exported as intermediate result, for subsequently determining the Dynamic Geographic attribute intermediate result in each place, for combining the quiet of each place
The credit scoring of state geographical attribute and Dynamic Geographic attribute map user obtains more comprehensive, accurate, reliable credit and comments
Estimate.
2) Dynamic Geographic attribute
Dynamic Geographic attribute is described using the type of the space-time migration model of user and motion track pattern, to two kinds of moulds
The classification of the extraction of formula and determining pattern illustrates.
By shown in Fig. 9 input data and intermediate result be used to form Dynamic Geographic attribute, and be based on credit mapping model
It is mapped to obtain user credit, input data and intermediate result include that n above-mentioned ties up geographical location sequence, n dimension place times
Sequence and m tie up " place-classification " list (or matrix).
2.1) motion track pattern
Referring to Figure 10, determine that motion track pattern ties up geographical location sequence, in one embodiment, Ke Yitong dependent on n
It crosses such mode and determines motion track pattern, the geographical location in the sequence of geographical location is based at least one partition of the scale
(such as in time scale and space scale) is multiple segmentations, and the geographical location in each segmentation corresponds to the movement of user
Track, the motion track on different scale have corresponded to motion track pattern of the user on different scale.Such as in residential quarters
Scale on, user may take a walk in cell, or travel to and fro between the supermarket shopping in cell, or various entertainments are carried out in cell
Activity etc. pre-establishes a series of motion track mode type table on different scales, passes through structure according to different application demands
Build the classification that grader stamps at least one dimension in motion track mode type table to the motion track pattern on different scale
Label.Such as the following table 2 example:
Table 2
2.2) space-time migration model
N is tieed up into locations and regions time series and m dimension " place-classification " lists combine, extracts user different
The different subsequences (such as the place places A- B- place C) migrated between place, extract frequently the subsequence extracted
The subsequence of pattern, that is, the frequency of occurrences meet preset condition (such as frequency of occurrences are most higher than frequency threshold or the frequency of occurrences
High predetermined quantity) subsequence, as user place rank space-time migration model.
In addition, space-time migration model (such as residence-hospital-residence or CBD between different types of place
Region-high-tech park-suburbs regions-CBD etc.) in, there is also certain while being frequent under multiple category division systems
Migration between the residence time in these space-time migration models and each place, place is lasted, is migrated by the subsequence of pattern
Starting combines with completion moment etc. is migrated, and just constitutes " multidimensional space-time migration series ".By structural classification device, by these
Multidimensional space-time migration series are mapped in " the space-time migration semantic classes " pre-established, are formed to user's living habit or life
The quantitative expression of pattern, for example, " often work to 9 points at night and then goes the amusement of bar street then to go home to sleep to 2:00 AM ",
Or " working in morning is often late and is all smooth traffic area on the way " etc., above-mentioned quantitative expression can be used for map user
Credit.
3) credit mapping is handled
Based on motion track pattern derived above and space-time migration model, by the Nonlinear Classifier built in advance and return
Return model, the first mapping processing carried out based on motion track pattern and space-time migration model, can obtain multiple dimensions, directly
Or reflect the associated credit feature of user's loan repayment capacity and refund wish indirectly.
Illustratively, including:The living habit of user, work habit, sports health custom (such as adhere to outdoor fortune daily
Dynamic, indoor timing body-building etc. weekly), quality of life and health status (such as whether having serious disease etc. recently), philosophy of life (such as
Stringent punctual, work drive foot, life are careless and sloppy freely etc.), (such as " working day lives in the areas CBD for income level and quality of the life
Domain and weekend return the suburbs villa live and leisure ", or " often aircraft weekend of going on business often goes to the beach both at home and abroad during work
And sea " etc.), the level of consumption etc..
Based on the credit feature of user for carrying out the first mapping and handling, then by compressive classification and regression model to
The credit feature at family carries out the second mapping and handles to obtain the final credit scoring result of user.
The above-mentioned credit mapping model based on Dynamic Geographic attribute is provided that be credit mapping block, in addition to can be independent
Using and export credit appraisal result outside, can also will reflect multiple dimensional characteristics of user's loan repayment capacity and refund wish as
Intermediate result exports, for subsequently determining the Dynamic Geographic attribute intermediate result in each place, for combining the static state in each place geographical
The credit scoring of attribute and Dynamic Geographic attribute map user obtains more comprehensive, accurate, reliable credit evaluation.
The device of determining credit provided in an embodiment of the present invention is illustrated again, referring to the determination credit shown in Figure 11
One optional structural schematic diagram of device, including:Data acquisition module 10, geographical attribute constructing module 20, credit map mould
Block 30, direct sequential density laminating module 40 and matrix elasticity partition-merge module 50, are combined position data and are merged
Processing determines that the processing procedure of user credit illustrates each module.
Amalgamation mode one
In conjunction with Fig. 4, data acquisition module 10, for (data source 1 to data source n) to obtain n geographical location from data source
Sequence, geographical location sequence include geographical location of the user residing for different time.
Geographical attribute constructing module 20, for clustering the geographical location in n geographical location sequence to obtain place;It is based on
The corresponding geographical location in each place and time determine the geographical attribute in each place;Geographical attribute based on each place and correspondence
Time builds geographical attribute sequence, for each geographical location sequence formed n intermediate result (including geographical attribute sequence 1 to
Geographical attribute sequence n).
Credit mapping block 30 is additionally operable to be based on n intermediate result, by the corresponding geography in each place in geographical attribute sequence
Attribute and time carry out the first mapping and handle to obtain the credit feature of at least one dimension of user;At least one based on user
The credit feature of a dimension carries out the second mapping and handles to obtain the credit of user.
Amalgamation mode two
Data acquisition module 10, which is used to correspond to from data source 1 to data source n, obtains geographical location sequence namely geographical location
Sequence 1 is to geographical location sequence n.
Direct sequential density laminating module 40, for will be identical at least in geographical location sequence 1 to geographical location sequence n when
Between corresponding geographical location be overlapped, obtain geographical location sequence to be resolved namely geographical location integrated data sequence.
Optionally, direct sequential density laminating module 40 is additionally operable in geographical location sequence to be resolved be to correspond to not
Weight is distributed in geographical location with data source, and the weight distributed is that at least one reliability and accuracy based on data source are true
It is fixed.
Geographical attribute constructing module 20, for clustering the geographical location in the integrated data sequence of geographical location to obtain ground
Point;The geographical attribute in each place is determined based on the corresponding geographical location in each place and time;Geographical attribute based on each place
And the corresponding time builds geographical attribute sequence, forms (one) intermediate result.
Credit mapping block 30 is additionally operable to be based on intermediate result, by the corresponding geographical category in each place in geographical attribute sequence
Property and the time carry out the first mapping handle to obtain the credit feature of at least one dimension of user;Based on at least one of user
The credit feature of dimension carries out the second mapping and handles to obtain the credit of user.
Amalgamation mode three
1) multidimensional data matrix-split
Matrix elasticity partition-merge module 50 corresponds to data source n for data source 1 and obtains geographical location sequence, i.e.,
Position sequence 1 is managed to geographical location sequence n.Multidimensional data matrix is built based on geographical location sequence 1 to geographical location sequence n.
Matrix elasticity partition-merge module 50 builds multidimensional data matrix in this way:Based at least two geography
Geographical location corresponds to the sequencing of time in position sequence, and the geographical location at least two geographical location sequences is carried out pair
Neat and filling processing, by treated, geographical location sequence builds multidimensional data matrix as row vector.
Matrix elasticity partition-merge module 50, each column vector in multidimensional data matrix, is based on each column vector for identification
The dimension for including effective geographical location, by the location data block and multidimensional location data that multidimensional data matrix-split is alternative splicing
Block identifies location data block and multidimensional location data block in this manner:It identifies in poly-dimensional block data there are one only having
The continuous column vector in effective geographical location of dimension is location data block;Identifying has at least two dimensions in poly-dimensional block data
Effective geographical location continuous column vector be multidimensional location data block.
2) static geographical attribute is extracted
Geographical attribute constructing module 20 replaces corresponding geographical location in geographical position sequence using the place after cluster, obtains
Place time series;To in the time series of place each place and each place corresponding time from least one dimension carry out
Semantic classes is classified, and is obtained each place in the time series of place and is corresponded to the classification of at least one dimension as static geographical attribute.
3) Dynamic Geographic attribute is extracted
Geographical attribute constructing module 20 is additionally operable to extract user based on place time series and static geographical sequence of attributes
Migration subsequence between different location, static geography sequence of attributes is static geographical attribute and correspondence based on each place
Time build to obtain;The migration subsequence that the extraction frequency of occurrences meets preset condition is that the space-time migration model of user is dynamic
Geographical attribute.
Optionally, geographical attribute constructing module 20 is additionally operable to the geographical location in the sequence of geographical location being based at least one
Partition of the scale is at least two segmentations;To the corresponding motion track pattern in geographical location in each segmentation from least one dimension into
Row classification obtains motion track pattern and corresponds to the classification of at least one dimension to be dynamic geographical attribute.
4) credit maps
Credit mapping block 30 is additionally operable to the geographical attribute and dynamic of the corresponding static state in each place in geographical attribute sequence
At least one geographical attribute carry out the first mapping processing, obtain the credit feature of the reflection at least one dimension of user;It will reflection
The credit feature of at least one dimension of user carries out the second mapping and handles to obtain the credit of user.
Optionally, credit mapping block 30 is additionally operable at least two geographical attribute sequences for pending first mapping processing
Row distribution weight, the corresponding weight of geographical attribute sequence are that at least one reliability and accuracy based on derived data source are true
Fixed, derived data source is by output for determining that geographical sequence of attributes uses the data source in geographical location.
In conjunction with Fig. 1 to determining that the realization method of the device of information in practical applications illustrates.
Realization method 1) mobile terminal
The device of determining credit provided in an embodiment of the present invention may be embodied as the mobile end with hardware configuration shown in Fig. 1
The method of above-mentioned determination credit is implemented in end by mobile terminal by running application program or software function module.
For example, can be provided as (including a series of using the software function module of the programming languages exploitation such as C/C++, Java
The instruction executed for processor), be embedded into the various mobile terminal Apps based on systems such as Android or iOS (such as it is micro-
Letter etc.), to directly use the computing resource (processor) of mobile terminal itself to obtain the geographical attribute of user, and based on geography
Attribute calculates the credit scoring of user, and periodically or non-periodically by various network communication modes by data, intermediate result
Or final result sends long-range server to, or locally preserved in mobile terminal.
Realization method 2) server end
The embodiment of the present invention can be provided is write as individual application software or large-scale soft based on programming languages such as C/C++, Java
Software function module (including a series of instruction executed for processor) in part system, runs on server end, will connect
Original geographical location, treated intermediate result or the final credit of the mobile terminal from single or numerous users received
Update is calculated as a result, being integrated with historical data, intermediate result or the result of credit scoring on server in scoring
Credit scoring as a result, the other applications or software function then run in export server end real-time or non real-time
Module uses, can also write service device client database or file stored.
Realization method 3) distributed credit evaluation platform
The embodiment of the present invention is also provided as the distributed parallel computing platform of multiple servers composition, carries interactive net
The network interface (Web) or other kinds user interface form the geographical location information used for personal, group or enterprise and excavate and believe
Use Evaluation Platform.Existing data packet batch can be uploaded to platform to obtain various result of calculations (among such as by user
As a result with the final result of credit scoring), can also by real-time data stream transmitting to this platform come calculate and update knot in real time
Fruit (such as final result of intermediate result and credit scoring).
Realization method 4) server-side application interface (API, Application Interface) and plug-in unit
The embodiment of the present invention can be provided as the API of server end, software development kit (SDK, Soft Development
Toolkit) or plug-in unit, call, and be embedded into types of applications program for other server-side application developers.
Realization method 5) mobile device client end AP I and plug-in unit
The embodiment of the present invention can be provided as API, SDK or plug-in unit of mobile device end, for other mobile terminal application programs
Developer calls, and is embedded into types of applications program.
Realization method 6) high in the clouds open service
It is excavated in the available geographical location information of the embodiment of the present invention and credit evaluation platform, the embodiment of the present invention may be used also
It is provided as API, SDK and the plug-in unit etc. of geographical location information excavation and credit evaluation platform, packing is packaged into for inside and outside enterprise
Personnel open the cloud service used.Or in a suitable form by various results (intermediate result and the final result of credit scoring)
It is illustrated in various terminals to show in equipment, be inquired for personal, group or enterprises and institutions.
In conclusion the embodiment of the present invention has the advantages that:
1) the user's geographic position data (unequal interval acquired in mobile (social activity) Apps such as wechat can be utilized
Time series) obtain the credit scoring of user, the coverage rate of data acquisition is high, can carry out credit to most user and comment
Point.
2) indirect (non-explicit) shopping or balance data has been used to determine the credit scoring of user as collage-credit data,
It, can be based on the big portion of user due to the most of the time gathering geographic position data for the life cycle that can cover user to user
The geographical attribute in the place residing for point life cycle is updated comprehensive evaluation to the credit of user, compared with merely with direct purchase
Object or balance data (data directly related with the finance of user, transaction) assessment user credit scoring scheme to
The evaluation of family credit is more accurate.Meanwhile can with using direct shopping or balance data to carrying out the scheme of credit scoring
It is used in combination, so as to support to solve the problems, such as reference using rich and varied data.
3) data volume bigger obtained by, frequency higher is closeer wider to the life time covering of user, to different crowd
Covering it is wider, help to realize general favour finance.
4) using the scheme for obtaining geographic position data from multi-data source, plurality of devices, the nothing of device and data source are supported
Seam is connected and merges, and on the one hand realizes and covers the multi-angle of user's life time to further increase the essence of credit scoring
On the other hand parasexuality and reliability have also further widened the boundary of crowd's covering, help to realize general favour finance.
It will be appreciated by those skilled in the art that:Realize that all or part of step of above method embodiment can pass through journey
Sequence instructs relevant hardware to complete, and program above-mentioned can be stored in a computer read/write memory medium, which exists
When execution, step including the steps of the foregoing method embodiments is executed;And storage medium above-mentioned includes:Flash memory device is deposited at random
Access to memory (RAM, Random Access Memory), read-only memory (ROM, Read-Only Memory), magnetic disc or
The various media that can store program code such as CD.
If alternatively, the above-mentioned integrated unit of the present invention is realized in the form of software function module and as independent product
Sale in use, can also be stored in a computer read/write memory medium.Based on this understanding, the present invention is implemented
The technical solution of example substantially in other words can be expressed in the form of software products the part that the relevant technologies contribute,
The computer software product is stored in a storage medium, including some instructions are used so that computer installation (can be with
It is personal computer, server or network equipment etc.) execute all or part of each embodiment the method for the present invention.
And storage medium above-mentioned includes:Flash memory device, RAM, ROM, magnetic disc or CD etc. are various can to store program code
Medium.
The above description is merely a specific embodiment, but scope of protection of the present invention is not limited thereto, any
Those familiar with the art in the technical scope disclosed by the present invention, can easily think of the change or the replacement, and should all contain
Lid is within protection scope of the present invention.Therefore, protection scope of the present invention should be based on the protection scope of the described claims.
Claims (18)
1. a kind of method of determining credit, which is characterized in that the method includes:
Geographical location sequence is obtained based on data source, the geographical location sequence includes geographical position of the user residing for different time
It sets;
It clusters the geographical location in the geographical location sequence to obtain place;
The geographical attribute in each place is determined based on the corresponding geographical location in each place and time;
Geographical attribute and corresponding time based on each place build geographical attribute sequence, and the geographical attribute sequence includes
Static geography sequence of attributes and/or Dynamic Geographic sequence of attributes;
The corresponding geographical attribute in each place in the geographical attribute sequence and time the first mapping of progress are handled to obtain described
The credit feature of at least one dimension of user;
The credit feature of at least one dimension based on the user carries out the second mapping and handles to obtain the credit of the user;
The geographical attribute that each place is determined based on the corresponding geographical location in each place and time, packet
It includes:
The label that the geographical location in geographical position sequence is replaced using the label in the place after cluster, obtains place time sequence
Row;To in the place time series each place and each place corresponding time from least one dimension carry out language
Adopted category classification obtains each place in the place time series and corresponds to the classification of at least one dimension to be the described of static state
Geographical attribute;And/or
It is at least two segmentations that geographical location in the sequence of geographical location, which is based at least one partition of the scale,;To in the segmentation
The corresponding motion track pattern in geographical location classify from least one dimension, obtain the motion track pattern correspond to
The classification of a few dimension is the dynamic geographical attribute.
2. according to the method described in claim 1, it is characterized in that, it is described based on data source obtain geographical location sequence, including:
It is corresponded to from least two data sources and obtains at least two geographical location sequences;
The corresponding geographical location of same time at least two geographical location sequences is overlapped, obtain it is to be resolved describedly
Manage position sequence.
3. according to the method described in claim 2, it is characterized in that, the method further includes:
It is the geographical location distribution weight of corresponding different data sources, the power distributed in the geographical location sequence to be resolved
Weight determines at least one the reliability, accuracy and sampling density in source based on the data.
4. according to the method described in claim 1, it is characterized in that, it is described based on data source obtain geographical location sequence, including:
Based on at least two geographical location sequence construct multidimensional data matrixes for corresponding to acquisition from least two data sources;
It identifies each column vector in the multidimensional data matrix, includes the dimension in effective geographical location based on each column vector,
By the location data block and multidimensional location data block that the multidimensional data matrix-split is alternative splicing.
5. according to the method described in claim 4, it is characterized in that,
At least two geographical location sequence construct multidimensional data matrixes based on from the correspondence acquisition of at least two data sources, packet
It includes:
The sequencing that the time is corresponded to based on geographical location in the sequence of at least two geographical location, by least two geographical positions
It sets the geographical location in sequence and carries out alignment and filling processing, it will be described in treated geographical location sequence builds as row vector
Multidimensional data matrix.
6. according to the method described in claim 4, it is characterized in that,
Described includes the dimension in effective geographical location based on each column vector, is alternately to spell by the multidimensional data matrix-split
The location data block and multidimensional location data block connect, including:
Identify that it is described fixed only to have the continuous column vector in effective geographical location there are one dimension in the multidimensional data matrix
Bit data block;
Identify that the continuous column vector in effective geographical location at least two dimensions in the multidimensional data matrix is described
Multidimensional location data block.
7. according to the method described in claim 1, it is characterized in that, the method further includes:
The user is extracted between the different places based on the place time series and static geographical sequence of attributes
Subsequence is migrated, the static geographical sequence of attributes is static geographical attribute and corresponding time structure based on each place
It builds to obtain;
The migration subsequence that the extraction frequency of occurrences meets preset condition is that the space-time migration model of the user is dynamic
The geographical attribute.
8. according to the method described in claim 1, it is characterized in that, described that each place in the geographical attribute sequence is corresponding
Geographical attribute and time carry out the first mapping and handle to obtain the credit feature of at least one dimension of the user, including:
At least one geographical attribute and dynamic geographical attribute by the corresponding static state in each place in the geographical attribute sequence into
The first mapping of row is handled, and obtains the credit feature for reflecting at least one dimension of the user.
9. according to the method described in claim 8, it is characterized in that, the method further includes:
At least two geographical attribute sequences for the pending first mapping processing distribute weight, the geographical attribute sequence
It is based on the reliability in derived data source, accuracy and determining, the source at least one sampling density to arrange corresponding weight
Data source is used to determine the data source that the geographical attribute sequence uses geographical location by output.
10. a kind of device of determining credit, which is characterized in that described device includes:
Data acquisition module, for obtaining geographical location sequence based on data source, the geographical location sequence includes user not
With the geographical location residing for the time;
Geographical attribute constructing module, for clustering the geographical location in the geographical location sequence to obtain place;
The geographical attribute constructing module is additionally operable to determine based on the corresponding geographical location in each place and time each
The geographical attribute in the place;Geographical attribute and corresponding time based on each place build geographical attribute sequence, described
Geographical attribute sequence includes static geographical sequence of attributes and/or Dynamic Geographic sequence of attributes;
Credit mapping block, for the corresponding geographical attribute in each place in the geographical attribute sequence and time to be carried out first
Mapping handles to obtain the credit feature of at least one dimension of the user;
The credit mapping block, the credit feature for being additionally operable at least one dimension based on the user carry out at the second mapping
Reason obtains the credit of the user;
The geographical attribute constructing module is also used for the place after clustering and replaces in the geographical location sequence accordingly
Geographical location obtains place time series;
To in the place time series each place and each place corresponding time from least one dimension into
Row semantic classes is classified, and obtains each place in the place time series and correspond to the classification of at least one dimension to be static
The geographical attribute;And/or
The geographical attribute constructing module is additionally operable to the geographical location in the geographical location sequence being based at least one scale
It is divided at least two segmentations;
Classify from least one dimension to the corresponding motion track pattern in geographical location in each segmentation, obtains described
The classification that motion track pattern corresponds at least one dimension is the dynamic geographical attribute.
11. device according to claim 10, which is characterized in that
The data acquisition module is additionally operable to correspond at least two geographical location sequences of acquisition from least two data sources;
Described device further includes:Direct sequential density laminating module, is used for same time at least two geographical location sequences
Corresponding geographical location is overlapped, and obtains the geographical location sequence to be resolved.
12. according to the devices described in claim 11, which is characterized in that
The direct sequential density laminating module is additionally operable in the geographical location sequence to be resolved be corresponding different data
Weight is distributed in the geographical location in source, and the weight distributed is the reliability, accuracy in source based on the data and uses density extremely
It is one of few to determine.
13. device according to claim 10, which is characterized in that
Described device further includes:Matrix elasticity partition-merge module, for based on from least two data sources correspond to obtain to
Few two geographical location sequence construct multidimensional data matrixes;
It identifies each column vector in the multidimensional data matrix, includes the dimension in effective geographical location based on each column vector,
By the location data block and multidimensional location data block that the multidimensional data matrix-split is alternative splicing.
14. device according to claim 13, which is characterized in that
The matrix elasticity partition-merge module is additionally operable to correspond to based on geographical location in the sequence of at least two geographical location
Geographical location at least two geographical location sequences is carried out alignment and filling is handled, after processing by the sequencing of time
Geographical location sequence build the multidimensional data matrix as row vector.
15. device according to claim 13, which is characterized in that
The matrix elasticity partition-merge module is additionally operable to identify in the multidimensional data matrix only tool having there are one dimension
The continuous column vector for imitating geographical location is the location data block;
Identify that the continuous column vector in effective geographical location at least two dimensions in the multidimensional data matrix is described
Multidimensional location data block.
16. device according to claim 10, which is characterized in that
The geographical attribute constructing module is additionally operable to extract institute based on the place time series and static geographical sequence of attributes
Migration subsequence of the user between the different places is stated, the static geographical sequence of attributes is based on the quiet of each place
State geographical attribute and corresponding time build to obtain;
The migration subsequence that the extraction frequency of occurrences meets preset condition is that the space-time migration model of the user is dynamic
The geographical attribute.
17. device according to claim 10, which is characterized in that
The credit mapping block is additionally operable to the geographical attribute of the corresponding static state in each place in the geographical attribute sequence and moves
At least one geographical attribute of state carries out the first mapping processing, obtains the credit feature for reflecting at least one dimension of the user.
18. device according to claim 17, which is characterized in that
The credit mapping block is additionally operable at least two geographical attribute sequences for the pending first mapping processing
Weight is distributed, the corresponding weight of the geographical attribute sequence is reliability, accuracy and sampling density based on derived data source
At least one determine, the derived data source for output be used for determine the geographical attribute sequence using geographical location data
Source.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610409746.7A CN106097095B (en) | 2016-06-08 | 2016-06-08 | Determine the method and device of credit |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610409746.7A CN106097095B (en) | 2016-06-08 | 2016-06-08 | Determine the method and device of credit |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106097095A CN106097095A (en) | 2016-11-09 |
CN106097095B true CN106097095B (en) | 2018-07-27 |
Family
ID=57227997
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610409746.7A Active CN106097095B (en) | 2016-06-08 | 2016-06-08 | Determine the method and device of credit |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106097095B (en) |
Families Citing this family (12)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106682427A (en) * | 2016-12-29 | 2017-05-17 | 平安科技(深圳)有限公司 | Personal health condition assessment method and device based position services |
CN108876076A (en) * | 2017-05-09 | 2018-11-23 | 中国移动通信集团广东有限公司 | The personal credit methods of marking and device of data based on instruction |
CN107818506A (en) * | 2017-09-30 | 2018-03-20 | 上海壹账通金融科技有限公司 | Electronic installation, credit risk control method and storage medium |
CN108846743B (en) * | 2018-06-12 | 2021-11-30 | 京东数字科技控股有限公司 | Method and apparatus for generating information |
CN109242671B (en) * | 2018-08-29 | 2020-09-29 | 厦门市七星通联科技有限公司 | Credit default prediction method and system based on multi-view incomplete data |
CN109978682A (en) * | 2019-03-28 | 2019-07-05 | 上海拍拍贷金融信息服务有限公司 | Credit-graded approach, device and computer storage medium |
CN111161042A (en) * | 2019-11-26 | 2020-05-15 | 深圳壹账通智能科技有限公司 | Personal risk assessment method, device, terminal and storage medium |
CN112017063B (en) * | 2020-07-15 | 2024-04-26 | 北京淇瑀信息科技有限公司 | Resource allocation method and device based on comprehensive risk score and electronic equipment |
CN112016791B (en) * | 2020-07-15 | 2024-04-26 | 北京淇瑀信息科技有限公司 | Resource allocation method and device and electronic equipment |
CN112907360B (en) * | 2021-03-25 | 2024-06-07 | 深圳前海微众银行股份有限公司 | Risk assessment method, apparatus, storage medium, and program product |
CN113626415B (en) * | 2021-08-27 | 2024-02-23 | 天元大数据信用管理有限公司 | Credit data output method, equipment and medium |
CN116563013A (en) * | 2023-05-19 | 2023-08-08 | 深圳百流科技有限公司 | Funds routing method, system and storage medium |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103593349A (en) * | 2012-08-14 | 2014-02-19 | 中国科学院沈阳自动化研究所 | Movement position analysis method in sense network environment |
CN104866969A (en) * | 2015-05-25 | 2015-08-26 | 百度在线网络技术(北京)有限公司 | Personal credit data processing method and device |
-
2016
- 2016-06-08 CN CN201610409746.7A patent/CN106097095B/en active Active
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103593349A (en) * | 2012-08-14 | 2014-02-19 | 中国科学院沈阳自动化研究所 | Movement position analysis method in sense network environment |
CN104866969A (en) * | 2015-05-25 | 2015-08-26 | 百度在线网络技术(北京)有限公司 | Personal credit data processing method and device |
Also Published As
Publication number | Publication date |
---|---|
CN106097095A (en) | 2016-11-09 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106097095B (en) | Determine the method and device of credit | |
Toch et al. | Analyzing large-scale human mobility data: a survey of machine learning methods and applications | |
Wang et al. | Unsupervised machine learning in urban studies: A systematic review of applications | |
CN106101222A (en) | The method for pushing of information and device | |
US9123259B2 (en) | Discovering functional groups of an area | |
EP3241370B1 (en) | Analyzing semantic places and related data from a plurality of location data reports | |
Couclelis | Cellular worlds: a framework for modeling micro—macro dynamics | |
Miller | What about people in geographic information science | |
Yang et al. | Hotel location evaluation: A combination of machine learning tools and web GIS | |
Li et al. | Statistical analysis of tourist flow in tourist spots based on big data platform and DA-HKRVM algorithms | |
CN109684330A (en) | User's portrait base construction method, device, computer equipment and storage medium | |
US20190242720A1 (en) | Systems and methods for constructing spatial activity zones | |
CN106570722A (en) | Intelligent recommendation system and intelligent recommendation method | |
Kim | Exploring the difference between ridership patterns of subway and taxi: Case study in Seoul | |
McKenzie et al. | Measuring urban regional similarity through mobility signatures | |
CN104679810A (en) | Computing Device For Generating Profiles Based On Mobile Device Data | |
Wang et al. | Measuring urban vibrancy of residential communities using big crowdsourced geotagged data | |
CN115408618A (en) | Interest point recommendation method based on social relationship fusion position dynamic popularity and geographic features | |
CN106649380A (en) | Hot spot recommendation method and system based on tag | |
Buscema et al. | A nonlinear, data-driven, ANNs-based approach to culture-led development policies in rural areas: The case of Gjakove and Peć districts, Western Kosovo | |
National Research Council et al. | IT roadmap to a geospatial future | |
Liu et al. | Mining method based on semantic trajectory frequent pattern | |
Encalada et al. | Mining big data for tourist hot spots: Geographical patterns of online footprints | |
CN115169718A (en) | Cellular automaton-based regional landslide risk dynamic prediction method and device | |
Efthymiou et al. | Modeling the impact of large-scale transportation infrastructure development on land cover |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |