CN107666649A - Personal property state evaluating method and device - Google Patents
Personal property state evaluating method and device Download PDFInfo
- Publication number
- CN107666649A CN107666649A CN201611246227.XA CN201611246227A CN107666649A CN 107666649 A CN107666649 A CN 107666649A CN 201611246227 A CN201611246227 A CN 201611246227A CN 107666649 A CN107666649 A CN 107666649A
- Authority
- CN
- China
- Prior art keywords
- user
- property
- personal property
- representation data
- seed
- 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
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W4/00—Services specially adapted for wireless communication networks; Facilities therefor
- H04W4/02—Services making use of location information
- H04W4/021—Services related to particular areas, e.g. point of interest [POI] services, venue services or geofences
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
-
- 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
- G06Q40/08—Insurance
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Business, Economics & Management (AREA)
- Physics & Mathematics (AREA)
- Finance (AREA)
- Accounting & Taxation (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Strategic Management (AREA)
- Evolutionary Computation (AREA)
- Computer Hardware Design (AREA)
- Development Economics (AREA)
- Economics (AREA)
- Marketing (AREA)
- Geometry (AREA)
- Technology Law (AREA)
- General Business, Economics & Management (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (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 personal property state evaluating method and device.The personal property appraisal procedure includes：The geographical position fence information of target area is obtained, the geographical position fence information includes houseclearing and corresponding targeted customer；Obtain the user representation data associated with the targeted customer；All targeted customers are divided into seed data collection and candidate data collection；The seed data collection includes at least one seed user, and the candidate data collection includes at least one candidate user；The property assessment models are trained using user's representation data of the seed user；According to user's representation data of the candidate user, the personal property state of the candidate user is assessed using the property assessment models, to export personal property condition evaluation results.In the personal property appraisal procedure, the personal property condition evaluation results of acquisition have higher accuracy, objectivity and reliability.
Description
Technical field
The present invention relates to technical field of information processing, more particularly to a kind of personal property state evaluating method and device.
Background technology
In financial institution household consumption management, financing planning management, Asset Allocation management and and investment pipe are provided to client
During the business such as reason, the personal property state of comprehensive assessment client is needed, and personal property state is arranged and analyzed, in time hair
The financial hidden danger of existing client, bad financing custom is corrected, improve the ability of resistance financial risks.Existing financial institution mainly utilizes
The finance data such as the condition of assets of client and consumption flowing water assesses personal property state, and assessment data source is single, causes individual
Property condition evaluation results accuracy rate is relatively low.Financial institution is based on the relatively low personal property condition evaluation results of accuracy rate to client
When service is provided, there is provided with its unmatched business of personal property state, financial risks may be caused.
The content of the invention
The technical problem to be solved in the present invention is, during for personal property state estimation in the prior art, assesses data
The deficiency that source is single and personal property condition evaluation results accuracy rate is relatively low, there is provided a kind of personal property state evaluating method and
Device.
The technical solution adopted for the present invention to solve the technical problems is：A kind of personal property state evaluating method, including：
The geographical position fence information of target area is obtained, the geographical position fence information is including houseclearing and correspondingly
Targeted customer；
Obtain the user representation data associated with the targeted customer；
All targeted customers are divided into seed data collection and candidate data collection；The seed data collection includes at least one
Seed user, the candidate data collection include at least one candidate user；
The property assessment models are trained using user's representation data of the seed user；
According to user's representation data of the candidate user, using the property assessment models to the candidate user
People's property state is assessed, to export personal property condition evaluation results.
Preferably, user's representation data using the seed user trains the property assessment models, including：
User's representation data of all seed users is classified using look-alike algorithms, obtains some sons
Representation data is shared corresponding to cluster and each sub-cluster；
The personal property state of each seed user is obtained, and calculates all seed users in each sub-cluster
Personal property average；
Representation data will be shared described in each sub-cluster and carries out logistic regression processing with the personal property average,
To obtain the property assessment models.
Preferably, user's representation data according to the candidate user, using the property assessment models to described
The personal property state of candidate user is assessed, to export personal property condition evaluation results, including：
The described common of user's representation data of the candidate user and each sub-cluster is calculated using similarity algorithm
There is the similarity of representation data；
Judge whether the similarity is more than default similar threshold value；
If so, then exported personal property average corresponding to sub-cluster as the personal property condition evaluation results.
Preferably, user's representation data includes the geographical location information based on location-based service, the geographical position letter
Breath includes the POI with time correlation connection.
Preferably, the geographical position fence information for obtaining target area, including：
House property medium platform and/or building property registration platform are crawled using web crawlers, to obtain the geographical position of target area
Put fence information.
The present invention also provides a kind of personal property state evaluation device, including：
Fence data obtaining module, for obtaining the geographical position fence information of target area, the geographical position fence
Information includes houseclearing and corresponding targeted customer；
Representation data acquisition module, for obtaining the user representation data associated with the targeted customer；
Data set division module, for all targeted customers to be divided into seed data collection and candidate data collection；The kind
Sub Data Set includes at least one seed user, and the candidate data collection includes at least one candidate user；
Assessment models training module, for training the property to assess mould using user's representation data of the seed user
Type；
Property state estimation module, for user's representation data according to the candidate user, assessed using the property
Model is assessed the personal property state of the candidate user, to export personal property condition evaluation results.
Preferably, the assessment models training module includes：
Representation data taxon, for being entered using look-alike algorithms to user's representation data of all seed users
Row classification, obtains and representation data is shared corresponding to some sub-clusters and each sub-cluster；
Property average calculation unit, for obtaining the personal property state of each seed user, and calculate each institute
State the personal property average of all seed users in sub-cluster；
Assessment models processing unit, for representation data and the personal property will to be shared described in each sub-cluster
Average carries out logistic regression processing, to obtain the property assessment models.
Preferably, the property state estimation module includes：
Similarity calculated, for calculated using similarity algorithm user's representation data of the candidate user with it is each
The similarity of the shared representation data of the sub-cluster；
Similarity-rough set unit, for judging whether the similarity is more than default similar threshold value；
Assessment result output unit, for if so, then using personal property average corresponding to sub-cluster as the personal wealth
Produce condition evaluation results output.
Preferably, user's representation data includes the geographical location information based on location-based service, the geographical position letter
Breath includes the POI with time correlation connection.
Preferably, the fence data obtaining module, for crawling house property medium platform and/or house property using web crawlers
Platform is registered, to obtain the geographical position fence information of target area.
The present invention has the following advantages that compared with prior art：Personal property appraisal procedure provided by the present invention and device
In, the geographical position fence information (including targeted customer) of target area is first obtained, and obtain the use associated with targeted customer
Family representation data；Targeted customer is divided into seed user and candidate user；Trained using user's representation data of seed user
Property assessment models, and user's representation data of candidate user is handled using the property assessment models trained, export
The personal property assessment result of candidate user.Mould is assessed using user's representation data training property of the seed user of target area
Type, personal property assessment is carried out to the candidate user of target area using the property assessment models trained so that personal property
Condition evaluation results have higher accuracy, objectivity and reliability.
Brief description of the drawings
Below in conjunction with drawings and Examples, the invention will be further described, in accompanying drawing：
Fig. 1 is a flow chart of personal property state evaluating method in the embodiment of the present invention 1；
Fig. 2 is a theory diagram of personal property state evaluation device in the embodiment of the present invention 2.
Embodiment
In order to which technical characteristic, purpose and the effect of the present invention is more clearly understood, now compares accompanying drawing and describe in detail
The embodiment of the present invention.
Embodiment 1
Fig. 1 shows a flow chart of personal property state evaluating method in the present embodiment.The personal property state estimation side
Method can be applied in the terminal of the financial institutions such as bank, insurance, for assessing the personal property state of any user.Such as Fig. 1 institutes
Show, the personal property state evaluating method comprises the following steps：
S10：The geographical position fence information of target area is obtained, geographical position fence information is including houseclearing and correspondingly
Targeted customer.
Wherein, target area can be any residential quarters.Houseclearing can be appointed in target area (residential quarters)
The information such as the house location in one house, house room number, house size, housing sale average price, house rent average price.Targeted customer can
To be housing ownership people corresponding to the houseclearing.In the present embodiment, the geographical position fence information of target area is obtained, with
Obtain the houseclearing in each house and corresponding targeted customer in any residential quarters, due to targeted customer live in it is same live
In residence cell, its personal property state has certain similitude, in order to the target determined based on geographical position fence information
User carries out personal property state estimation.Residential quarters corresponding to the target area are preferably the higher house of housing sale average price
Cell, housing sale average price superelevation, its corresponding house householder (i.e. targeted customer) should have corresponding personal property state.
Specifically, step S10 is specifically included：House property medium platform and/or building property registration platform are crawled using web crawlers,
To obtain the geographical position fence information of target area.
Specifically, web crawlers is captured the room in house property medium platform and/or building property registration platform by preset rules automatically
Room information, and obtain the targeted customer associated with houseclearing, the ground using houseclearing and targeted customer as target area
Manage positional information output.It is to be appreciated that any room in target area is stored with house property medium platform and/or building property registration platform
The houseclearing in room and targeted customer, target area is crawled from house property medium platform and/or building property registration platform using web crawlers
The geographical position fence information in domain, crawls that data content is clear and definite, and processing speed is very fast.
In the present embodiment, targeted customer is housing ownership people corresponding with any houseclearing of target area, same
The personal property state of the targeted customer of target area has certain similarity.It is to be appreciated that the house with target area
Other same or analogous regions of sale average price can also be used as same target area, to expand targeted customer's model of target area
Enclose.Targeted customer based on target area carries out personal property assessment, can improve personal property state estimation to a certain extent
As a result accuracy and reliability.
S20：Obtain the user representation data associated with targeted customer.
User's representation data (i.e. Persona data) is the virtual representations of real user, is built upon a system and truly counts
According to targeted customer's model on (Marketing Data/Usability Data).The depository financial institutions such as current bank, insurance
User's representation data of the targeted customer of storage includes but is not limited to address name, identification feature, photo, contact method, family
Front yard address, office space, occupation and income etc..In the present embodiment, obtain in user's representation data associated with targeted customer,
User is associated with the houseclearing in geography information fence information and targeted customer corresponding to each user's representation data, so that
User's representation data of each targeted customer includes the houseclearing of target area, so as to based on the mesh related to houseclearing
User's representation data of mark user is handled,
Specifically, user's representation data includes the geographical location information based on location-based service, geographical location information include with
The POI of time correlation connection.
By taking targeted customer's geographical location information of one day as an example, the geographical location information includes 0：00—24：00 POI
Information, each POI are used to indicate a bit in electronic map, including the information such as POI points title, longitude and latitude.Pass through
To the targeted customer, geographical location information daily within a period of time is handled, it may be determined that the family of the targeted customer lives
Location, office space, commuter time, the amusement often gone, consumption, body-building etc..It is to be appreciated that user's representation data can be with
Including consumption feature, investment characteristics or other influences personal property assessment factor feature.It is to be appreciated that if targeted customer is frequent
Property come in and go out top-grade consumption place, or have the personal property of the information, the then targeted customer such as wholesale investment record in financial institution
Relatively low, the personal property assessment result that it is obtained is higher, to ensure the accuracy of personal property assessment.Ground based on location-based service
Reason positional information is the daily life track of targeted customer, has objectivity, and personal property shape is carried out based on geographical location information
State is assessed, and can be advantageous to improve the objectivity and accuracy of personal property condition evaluation results, be avoided only in accordance with targeted customer certainly
The problem of subjectivity that the information of main offer is carried out caused by personal property assessment is strong, and assessment result accuracy is low occurs.
Wherein, it is by telecommunications mobile operator based on location-based service (Location Based Service, abbreviation LBS)
Radio communication network (such as GSM nets, CDMA nets) or outside positioning method (such as GPS) obtain mobile terminal (i.e. targeted customer)
Positional information (geographical coordinate, or geodetic coordinates), GIS-Geographic Information System (Geographic Information System,
Abbreviation GIS) platform support under, a kind of value-added service of respective service is provided for targeted customer.All in all, LBS is by moving
Communication network and computer network are combined into, and are realized and interacted by gateway between two networks.Mobile terminal passes through mobile logical
Communication network sends request, and LBS service platform is given by gateway passes；LBS service platform is used according to targeted customer's request and target
Family current location is handled, and result is returned into targeted customer by gateway.POI (Point Of Interest, it is impromptu
Interest point or information point), including the data such as title, type, longitude, latitude, so that POI can be presented on the electronic map, with sign
The location informations such as some terrestrial reference, sight spot on electronic map.
In the present embodiment, the mobile terminal based on location-based service is smart mobile phone, by opening the positioning on smart mobile phone
Function, so that LBS service platform obtains the geographical location information of smart mobile phone in real time, so as to understand the mesh for carrying the smart mobile phone
Mark the geographical location information of user.Geographical location information include with time correlation connection POI in time include the date and
At the moment, targeted customer's POI residing at any one time can be appreciated that by the geographical location information.It is it is to be appreciated that geographical
Positional information is associated with the ID of targeted customer, ID be used for identify unique identification user, can be identification card number or
Cell-phone number.
S30：All targeted customers are divided into seed data collection and candidate data collection；Seed data collection includes at least one
Seed user, candidate data collection include at least one candidate user.
In the present embodiment, to all targeted customers in target area, whether foundation, which carried out personal property assessment and had, is commented
All targeted customers in target area are divided into seed data collection and candidate data by the personal property state estimated as division condition
Collection.Wherein, seed data concentrates each seed user to have the personal property state after assessing.Candidate data concentrates each candidate
User is without the personal property state after assessing.
S40：Property assessment models are trained using user's representation data of seed user.
Because seed data concentrates the personal property state that all seed users are respectively provided with after assessing, and each seed user
Corresponding user's representation data is respectively provided with, user's representation data includes but is not limited to address name, identification feature, shone
Piece, contact method, home address, office space, occupation and income etc., in addition to embody the base of targeted customer's daily life track
In the geographic position data of location-based service.Commonness Analysis is carried out to user's representation data of all seed users, seed is obtained and uses
Incidence relation between the representation data and personal property state at family, to train property assessment models.
It is to be appreciated that each seed user is the targeted customer of target area, user's portrait number of seed user is utilized
According to property assessment models are trained, the accuracy and reliability of personal property condition evaluation results can be improved to a certain extent.And
And user's representation data includes being used for the geographic position data for embodying seed user daily life track, has objectivity, is based on
Geographical location information trains property assessment models, can be advantageous to improve the objectivity of personal property condition evaluation results and accurate
Property.
Further, step S40 comprises the following steps：
S41：User's representation data of all seed users is classified using look-alike algorithms, obtains some sons
Representation data is shared corresponding to cluster and each sub-cluster.
Wherein, Look-alike, i.e., similar crowd's extension, is that one kind is based on existing user/device id, passes through certain calculation
Method assessment models, find the technology for the similar crowd for more possessing potential relevance.The present embodiment uses look-alike to calculate
User's representation data of seed user is used in method as positive sample, train classification models to obtain shared representation data, in order to
User's representation data of candidate user is used to be screened for negative sample by disaggregated model.
Specifically, user's representation data of all seed users adopt in assorting process using look-alike algorithms
Use based on PU-Learning (Learning from Positive and Unlabled Example, i.e. positive example and without mark
Note sample learning) sorting technique classified, assorting process is simple and convenient, can effectively reduce the preparation work of manual sort
Amount, improve nicety of grading.It is to be appreciated that user's representation data of all seed users is carried out using look-alike algorithms
There is identical to share representation data for classification, each sub-cluster of acquisition, be that can be used for the association for assessing personal property state special
Sign.
Due to user's representation data of each seed user include embodying targeted customer's daily life track based on position
The geographic position data of service, classification acquisition is carried out to user's representation data of all seed users using look-alike algorithms
Each sub-cluster in shared representation data it is associated with the geographic position data based on location-based service, there is objectivity and can
By property.
S42：The personal property state of each seed user is obtained, and calculates of all seed users in each sub-cluster
People's property average.
Because seed data concentrates each seed user to have the personal property state after assessing, then using look-alike
Seed user in each sub-cluster that algorithm is classified and got to user's representation data of all seed users also has
There is the personal property state after assessing.In the present embodiment, the personal property average of all seed users in each sub-cluster is calculated,
Property assessment models are built using personal property average.
S43：Shared representation data in each sub-cluster and personal property average are subjected to logistic regression processing, to obtain
Property assessment models.
In the present embodiment, the personal property average of the shared representation data of each sub-cluster and the sub-cluster is used into logic
Regression algorithm carries out logistic regression processing, to obtain property assessment models.In the property assessment models, the shared picture in sub-cluster
As data and personal property average establish mapping relations.Wherein, representation data and the geographical position number based on location-based service are shared
According to being associated, there is objectivity and reliability, the property assessment models formed it into have objectivity and reliability.
Wherein, logistic regression (Logistic Regression) is the more commonly used machine learning method of current industry, is used
In the possibility for estimating certain things.Logistic regression (Logistic Regression) is one by logistic equation normalizings
Linear regression after change.In logistic regression (Logistic Regression), if it is { x, y } to set sample, y is 0 or 1, table
Show positive class or negative class, x is the sampling feature vectors of our m dimensions.So this sample x belongs to positive class, that is, y=1
" probability " can be represented by following logical function：
Wherein, θ is model parameter, that is, regression coefficient, σ are sigmoid functions.Actually this function is by following
Logarithm probability (namely x belongs to the logarithm of the ratio of the possibility of positive class and the possibility of negative class) conversion obtain：
S50：According to user's representation data of candidate user, the personal property shape using property assessment models to candidate user
State is assessed, to export personal property condition evaluation results.
In the property assessment models perfected, user's representation data is set to establish mapping relations with personal property state.For
Any candidate user without the personal property state assessed, user's representation data input of candidate user need to only be trained
Property assessment models carry out personal property state estimation, you can output personal property condition evaluation results, evaluation process is simple
It is convenient, and the personal property condition evaluation results exported have objectivity and accuracy, and evaluation process is simple to operate.
Further, step S50 comprises the following steps：
S51：The user's representation data and the shared portrait number of each sub-cluster of candidate user are calculated using similarity algorithm
According to similarity.
In the present embodiment, user's representation data of candidate user and each sub-cluster are calculated using text similarity measurement algorithm
The similarity of shared representation data.Calculating similarity using text similarity measurement algorithm includes following process：First, to candidate user
User's representation data segmented, go stop words etc. pre-process.Then, text spy is carried out based on TF-IDF or other weights
Sign extraction and weighting.Finally, cosine value calculating is carried out using vector space model, so that the user of candidate user is calculated
The similarity of representation data and the shared representation data of each sub-cluster.Wherein, TF (Term frequency, i.e. keyword word
Frequently), the frequency that keyword occurs in an article is referred to；IDF (Inverse document frequency, i.e., reverse text
Frequency), it is the index for weighing keyword weight.Phase recency is calculated using text similarity measurement algorithm, there is calculating process letter
Singly, the advantages of calculating speed is very fast.It is to be appreciated that the text similarity measurement algorithm based on semantic similarity, base can also be used
Handled in similarity algorithms such as the Chinese fuzzy search algorithms of pinyin similarity.
S52：Judge whether similarity is more than default similar threshold value.
Wherein, similar threshold value is pre-set, can be certainly for judging that candidate user belongs to the numerical value of any subset group
Main setting.In the present embodiment, similar threshold value is set to 70%.That is user's representation data of candidate user and the shared picture of a sub-cluster
When being more than similar threshold value (70%) as the similarity of data, then it is assumed that candidate user can belong to the sub-cluster.
S53：If so, then exported personal property average corresponding to sub-cluster as personal property condition evaluation results.
It is to be appreciated that if the similarity of user's representation data of candidate user and the shared representation data of a sub-cluster is big
When similar threshold value, it is believed that candidate user can belong to the sub-cluster, regard personal property average corresponding to the sub-cluster as this
The personal property condition evaluation results output of candidate user.In the present embodiment, the personal property state estimation of any candidate user
As a result it is associated with its user's representation data, and user's representation data includes the geographical location information correlation based on location-based service
Connection, has objectivity and reliability.
In the personal property appraisal procedure that the present embodiment is provided, the geographical position fence information of target area is first obtained
(including targeted customer), and obtain the user representation data associated with targeted customer；Targeted customer is divided into seed user
And candidate user；Property assessment models are trained using user's representation data of seed user, and assessed using the property trained
Model is handled user's representation data of candidate user, exports the personal property assessment result of candidate user.Using target
User's representation data training property assessment models of the seed user in region, using the property assessment models trained to target area
The candidate user in domain carries out personal property assessment, and evaluation process is simple and convenient, and the personal property condition evaluation results tool exported
There are higher accuracy, objectivity and reliability.
Embodiment 2
Fig. 2 shows a flow chart of personal property state evaluation device in the present embodiment.The personal property state estimation fills
Putting can apply in the financial institutions such as bank, insurance, for assessing the personal property state of any user.As shown in Fig. 2 this
People's property state evaluation device include fence data obtaining module 10, representation data acquisition module 20, data set division module 30,
Assessment models training module 40 and property state estimation module 50.
Fence data obtaining module 10, for obtaining the geographical position fence information of target area, geographical position fence letter
Breath includes houseclearing and corresponding targeted customer.
Wherein, target area can be any residential quarters.Houseclearing can be appointed in target area (residential quarters)
The information such as the house location in one house, house room number, house size, housing sale average price, house rent average price.Targeted customer can
To be housing ownership people corresponding to the houseclearing.In the present embodiment, the geographical position fence information of target area is obtained, with
Obtain the houseclearing in each house and corresponding targeted customer in any residential quarters, due to targeted customer live in it is same live
In residence cell, its personal property state has certain similitude, in order to the target determined based on geographical position fence information
User carries out personal property state estimation.Residential quarters corresponding to the target area are preferably the higher house of housing sale average price
Cell, housing sale average price superelevation, its corresponding house householder (i.e. targeted customer) should have corresponding personal property state.
Specifically, fence data obtaining module 10, for using, web crawlers crawls house property medium platform and/or house property is stepped on
Platform is remembered, to obtain the geographical position fence information of target area.
Specifically, web crawlers is captured the room in house property medium platform and/or building property registration platform by preset rules automatically
Room information, and obtain the targeted customer associated with houseclearing, the ground using houseclearing and targeted customer as target area
Manage positional information output.It is to be appreciated that any room in target area is stored with house property medium platform and/or building property registration platform
The houseclearing in room and targeted customer, target area is crawled from house property medium platform and/or building property registration platform using web crawlers
The geographical position fence information in domain, crawls that data content is clear and definite, and processing speed is very fast.
In the present embodiment, targeted customer is housing ownership people corresponding with any houseclearing of target area, same
The personal property state of the targeted customer of target area has certain similarity.It is to be appreciated that the house with target area
Other same or analogous regions of sale average price can also be used as same target area, to expand targeted customer's model of target area
Enclose.Targeted customer based on target area carries out personal property assessment, can improve personal property state estimation to a certain extent
As a result accuracy and reliability.
Representation data acquisition module 20, for obtaining the user representation data associated with targeted customer.
User's representation data (i.e. Persona data) is the virtual representations of real user, is built upon a system and truly counts
According to targeted customer's model on (Marketing Data/Usability Data).The depository financial institutions such as current bank, insurance
User's representation data of the targeted customer of storage includes but is not limited to address name, identification feature, photo, contact method, family
Front yard address, office space, occupation and income etc..In the present embodiment, obtain in user's representation data associated with targeted customer,
User is associated with the houseclearing in geography information fence information and targeted customer corresponding to each user's representation data, so that
User's representation data of each targeted customer includes the houseclearing of target area, so as to based on the mesh related to houseclearing
User's representation data of mark user is handled,
Specifically, user's representation data includes the geographical location information based on location-based service, geographical location information include with
The POI of time correlation connection.
By taking targeted customer's geographical location information of one day as an example, the geographical location information includes 0：00—24：00 POI
Information, each POI are used to indicate a bit in electronic map, including the information such as POI points title, longitude and latitude.Pass through
To the targeted customer, geographical location information daily within a period of time is handled, it may be determined that the family of the targeted customer lives
Location, office space, commuter time, the amusement often gone, consumption, body-building etc..It is to be appreciated that user's representation data can be with
Including consumption feature, investment characteristics or other influences personal property assessment factor feature.It is to be appreciated that if targeted customer is frequent
Property come in and go out top-grade consumption place, or have the personal property of the information, the then targeted customer such as wholesale investment record in financial institution
Relatively low, the personal property assessment result that it is obtained is higher, to ensure the accuracy of personal property assessment.Ground based on location-based service
Reason positional information is the daily life track of targeted customer, has objectivity, and personal property shape is carried out based on geographical location information
State is assessed, and can be advantageous to improve the objectivity and accuracy of personal property condition evaluation results, be avoided only in accordance with targeted customer certainly
The problem of subjectivity that the information of main offer is carried out caused by personal property assessment is strong, and assessment result accuracy is low occurs.
Wherein, it is by telecommunications mobile operator based on location-based service (Location Based Service, abbreviation LBS)
Radio communication network (such as GSM nets, CDMA nets) or outside positioning method (such as GPS) obtain mobile terminal (i.e. targeted customer)
Positional information (geographical coordinate, or geodetic coordinates), GIS-Geographic Information System (Geographic Information System,
Abbreviation GIS) platform support under, a kind of value-added service of respective service is provided for targeted customer.All in all, LBS is by moving
Communication network and computer network are combined into, and are realized and interacted by gateway between two networks.Mobile terminal passes through mobile logical
Communication network sends request, and LBS service platform is given by gateway passes；LBS service platform is used according to targeted customer's request and target
Family current location is handled, and result is returned into targeted customer by gateway.POI (Point Of Interest, it is impromptu
Interest point or information point), including the data such as title, type, longitude, latitude, so that POI can be presented on the electronic map, with sign
The location informations such as some terrestrial reference, sight spot on electronic map.
In the present embodiment, the mobile terminal based on location-based service is smart mobile phone, by opening the positioning on smart mobile phone
Function, so that LBS service platform obtains the geographical location information of smart mobile phone in real time, so as to understand the mesh for carrying the smart mobile phone
Mark the geographical location information of user.Geographical location information include with time correlation connection POI in time include the date and
At the moment, targeted customer's POI residing at any one time can be appreciated that by the geographical location information.It is it is to be appreciated that geographical
Positional information is associated with the ID of targeted customer, ID be used for identify unique identification user, can be identification card number or
Cell-phone number.
Data set division module 30, for all targeted customers to be divided into seed data collection and candidate data collection；Seed
Data set includes at least one seed user, and candidate data collection includes at least one candidate user.
In the present embodiment, to all targeted customers in target area, whether foundation, which carried out personal property assessment and had, is commented
All targeted customers in target area are divided into seed data collection and candidate data by the personal property state estimated as division condition
Collection.Wherein, seed data concentrates each seed user to have the personal property state after assessing.Candidate data concentrates each candidate
User is without the personal property state after assessing.
Assessment models training module 40, for training property assessment models using user's representation data of seed user.
Because seed data concentrates the personal property state that all seed users are respectively provided with after assessing, and each seed user
Corresponding user's representation data is respectively provided with, user's representation data includes but is not limited to address name, identification feature, shone
Piece, contact method, home address, office space, occupation and income etc., in addition to embody the base of targeted customer's daily life track
In the geographic position data of location-based service.Commonness Analysis is carried out to user's representation data of all seed users, seed is obtained and uses
Incidence relation between the representation data and personal property state at family, to train property assessment models.
It is to be appreciated that each seed user is the targeted customer of target area, user's portrait number of seed user is utilized
According to property assessment models are trained, the accuracy and reliability of personal property condition evaluation results can be improved to a certain extent.And
And user's representation data includes being used for the geographic position data for embodying seed user daily life track, has objectivity, is based on
Geographical location information trains property assessment models, can be advantageous to improve the objectivity of personal property condition evaluation results and accurate
Property.
Further, assessment models training module 40 specifically includes representation data taxon 41, property mean value computation list
Member 42 and assessment models processing unit 43.
Representation data taxon 41, for user's representation data using look-alike algorithms to all seed users
Classified, obtain and representation data is shared corresponding to some sub-clusters and each sub-cluster.
Wherein, Look-alike, i.e., similar crowd's extension, is that one kind is based on existing user/device id, passes through certain calculation
Method assessment models, find the technology for the similar crowd for more possessing potential relevance.The present embodiment uses look-alike to calculate
User's representation data of seed user is used in method as positive sample, train classification models to obtain shared representation data, in order to
User's representation data of candidate user is used to be screened for negative sample by disaggregated model.
Specifically, user's representation data of all seed users adopt in assorting process using look-alike algorithms
Use based on PU-Learning (Learning from Positive and Unlabled Example, positive example and unmarked
Sample learning) sorting technique classified, assorting process is simple and convenient, can effectively reduce the preparation work amount of manual sort,
Improve nicety of grading.It is to be appreciated that user's representation data of all seed users is divided using look-alike algorithms
There is identical to share representation data for class, each sub-cluster of acquisition, be the linked character that can be used for assessing personal property state.
Due to user's representation data of each seed user include embodying targeted customer's daily life track based on position
The geographic position data of service, classification acquisition is carried out to user's representation data of all seed users using look-alike algorithms
Each sub-cluster in shared representation data it is associated with the geographic position data based on location-based service, there is objectivity and can
By property.
Property average calculation unit 42, for obtaining the personal property state of each seed user, and calculate each subset
The personal property average of all seed users in group.
Because seed data concentrates each seed user to have the personal property state after assessing, then using look-alike
Seed user in each sub-cluster that algorithm is classified and got to user's representation data of all seed users also has
There is the personal property state after assessing.In the present embodiment, the personal property average of all seed users in each sub-cluster is calculated,
Property assessment models are built using personal property average.
Assessment models processing unit 43, for the shared representation data in each sub-cluster and personal property average to be carried out
Logistic regression processing, to obtain property assessment models.
In the present embodiment, the personal property average of the shared representation data of each sub-cluster and the sub-cluster is used into logic
Regression algorithm carries out logistic regression processing, to obtain property assessment models.In the property assessment models, the shared picture in sub-cluster
As data and personal property average establish mapping relations.Wherein, representation data and the geographical position number based on location-based service are shared
According to being associated, there is objectivity and reliability, the property assessment models formed it into have objectivity and reliability.
Wherein, logistic regression (Logistic Regression) is the more commonly used machine learning method of current industry, is used
In the possibility for estimating certain things.Logistic regression (Logistic Regression) is one by logistic equation normalizings
Linear regression after change.In logistic regression (Logistic Regression), if it is { x, y } to set sample, y is 0 or 1, table
Show positive class or negative class, x is the sampling feature vectors of our m dimensions.So this sample x belongs to positive class, that is, y=1
" probability " can be represented by following logical function：
Wherein, θ is model parameter, that is, regression coefficient, σ are sigmoid functions.Actually this function is by following
Logarithm probability (namely x belongs to the logarithm of the ratio of the possibility of positive class and the possibility of negative class) conversion obtain：
Property state estimation module 50, for user's representation data according to candidate user, utilize property assessment models pair
The personal property state of candidate user is assessed, to export personal property condition evaluation results.
In the property assessment models perfected, user's representation data is set to establish mapping relations with personal property state.For
Any candidate user without the personal property state assessed, user's representation data input of candidate user need to only be trained
Property assessment models carry out personal property state estimation, you can output personal property condition evaluation results, evaluation process is simple
It is convenient, and the personal property condition evaluation results exported have objectivity and accuracy, and evaluation process is simple to operate.
Further, property state estimation module 50 specifically includes similarity calculated 51, similarity-rough set unit 52
With assessment result output unit 53.
Similarity calculated 51, for user's representation data that candidate user is calculated using similarity algorithm and each son
The similarity of the shared representation data of cluster.
In the present embodiment, user's representation data of candidate user and each sub-cluster are calculated using text similarity measurement algorithm
The similarity of shared representation data.Calculating similarity using text similarity measurement algorithm includes following process：First, to candidate user
User's representation data segmented, go stop words etc. pre-process.Then, text spy is carried out based on TF-IDF or other weights
Sign extraction and weighting.Finally, cosine value calculating is carried out using vector space model, so that the user of candidate user is calculated
The similarity of representation data and the shared representation data of each sub-cluster.Wherein, TF (Term frequency, i.e. keyword word
Frequently), the frequency that keyword occurs in an article is referred to；IDF (Inverse document frequency, i.e., reverse text
Frequency), it is the index for weighing keyword weight.Phase recency is calculated using text similarity measurement algorithm, there is calculating process letter
Singly, the advantages of calculating speed is very fast.It is to be appreciated that the text similarity measurement algorithm based on semantic similarity, base can also be used
Handled in similarity algorithms such as the Chinese fuzzy search algorithms of pinyin similarity.
Similarity-rough set unit 52, for judging whether similarity is more than default similar threshold value.
Wherein, similar threshold value is pre-set, can be certainly for judging that candidate user belongs to the numerical value of any subset group
Main setting.In the present embodiment, similar threshold value is set to 70%.That is user's representation data of candidate user and the shared picture of a sub-cluster
When being more than similar threshold value (70%) as the similarity of data, then it is assumed that candidate user can belong to the sub-cluster.
Assessment result output unit 53, for if so, then using personal property average corresponding to sub-cluster as personal property
Condition evaluation results export.
It is to be appreciated that if the similarity of user's representation data of candidate user and the shared representation data of a sub-cluster is big
When similar threshold value, it is believed that candidate user can belong to the sub-cluster, regard personal property average corresponding to the sub-cluster as this
The personal property condition evaluation results output of candidate user.In the present embodiment, the personal property state estimation of any candidate user
As a result it is associated with its user's representation data, and user's representation data includes the geographical location information correlation based on location-based service
Connection, has objectivity and reliability.
In the personal property apparatus for evaluating that the present embodiment is provided, the geographical position fence information of target area is first obtained
(including targeted customer), and obtain the user representation data associated with targeted customer；Targeted customer is divided into seed user
And candidate user；Property assessment models are trained using user's representation data of seed user, and assessed using the property trained
Model is handled user's representation data of candidate user, exports the personal property assessment result of candidate user.Using target
User's representation data training property assessment models of the seed user in region, using the property assessment models trained to target area
The candidate user in domain carries out personal property assessment, and evaluation process is simple and convenient, and the personal property condition evaluation results tool exported
There are higher accuracy, objectivity and reliability.
The present invention is illustrated by several specific embodiments, it will be appreciated by those skilled in the art that, do not departing from
In the case of the scope of the invention, various conversion and equivalent substitute can also be carried out to the present invention.In addition, it is directed to particular condition or tool
Body situation, various modifications can be made to the present invention, without departing from the scope of the present invention.Therefore, the present invention is not limited to disclosed
Specific embodiment, and whole embodiments for falling within the scope of the appended claims should be included.
Claims (10)
- A kind of 1. personal property state evaluating method, it is characterised in that including：The geographical position fence information of target area is obtained, the geographical position fence information includes houseclearing and corresponding mesh Mark user；Obtain the user representation data associated with the targeted customer；All targeted customers are divided into seed data collection and candidate data collection；The seed data collection includes at least one seed User, the candidate data collection include at least one candidate user；The property assessment models are trained using user's representation data of the seed user；According to user's representation data of the candidate user, the personal wealth using the property assessment models to the candidate user Occurrence state is assessed, to export personal property condition evaluation results.
- 2. personal property state evaluating method according to claim 1, it is characterised in that described to utilize the seed user User's representation data train the property assessment models, including：User's representation data of all seed users is classified using look-alike algorithms, obtains some sub-clusters And representation data is shared corresponding to each sub-cluster；The personal property state of each seed user is obtained, and calculates of all seed users in each sub-cluster People's property average；Representation data will be shared described in each sub-cluster and carries out logistic regression processing with the personal property average, to obtain Take the property assessment models.
- 3. personal property state evaluating method according to claim 2, it is characterised in that described according to the candidate user User's representation data, the personal property state of the candidate user is assessed using the property assessment models, with defeated Go out personal property condition evaluation results, including：The user's representation data and the shared picture of each sub-cluster of the candidate user are calculated using similarity algorithm As the similarity of data；Judge whether the similarity is more than default similar threshold value；If so, then exported personal property average corresponding to sub-cluster as the personal property condition evaluation results.
- 4. personal property state evaluating method according to claim 1, it is characterised in that user's representation data includes Geographical location information based on location-based service, the geographical location information include the POI with time correlation connection.
- 5. personal property state evaluating method according to claim 1, it is characterised in that the ground for obtaining target area Position fence information is managed, including：House property medium platform and/or building property registration platform are crawled using web crawlers, enclosed with obtaining the geographical position of target area Column information.
- A kind of 6. personal property state evaluation device, it is characterised in that including：Fence data obtaining module, for obtaining the geographical position fence information of target area, the geographical position fence information Including houseclearing and corresponding targeted customer；Representation data acquisition module, for obtaining the user representation data associated with the targeted customer；Data set division module, for all targeted customers to be divided into seed data collection and candidate data collection；The seed number Include at least one seed user according to collection, the candidate data collection includes at least one candidate user；Assessment models training module, for training the property assessment models using user's representation data of the seed user；Property state estimation module, for user's representation data according to the candidate user, utilize the property assessment models The personal property state of the candidate user is assessed, to export personal property condition evaluation results.
- 7. personal property state evaluation device according to claim 6, it is characterised in that the assessment models training module Including：Representation data taxon, for being divided using look-alike algorithms user's representation data of all seed users Class, obtain and representation data is shared corresponding to some sub-clusters and each sub-cluster；Property average calculation unit, for obtaining the personal property state of each seed user, and calculate each son The personal property average of all seed users in cluster；Assessment models processing unit, for representation data and the personal property average will to be shared described in each sub-cluster Logistic regression processing is carried out, to obtain the property assessment models.
- 8. personal property state evaluation device according to claim 7, it is characterised in that the property state estimation module Including：Similarity calculated, for calculated using similarity algorithm user's representation data of the candidate user with it is each described The similarity of the shared representation data of sub-cluster；Similarity-rough set unit, for judging whether the similarity is more than default similar threshold value；Assessment result output unit, for if so, then using personal property average corresponding to sub-cluster as the personal property shape State assessment result exports.
- 9. personal property state evaluation device according to claim 6, it is characterised in that user's representation data includes Geographical location information based on location-based service, the geographical location information include the POI with time correlation connection.
- 10. personal property state evaluation device according to claim 6, it is characterised in that the fence acquisition of information mould Block, for crawling house property medium platform and/or building property registration platform using web crawlers, to obtain the geographical position of target area Fence information.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611246227.XA CN107666649A (en) | 2016-12-29 | 2016-12-29 | Personal property state evaluating method and device |
| PCT/CN2017/076463 WO2018120425A1 (en) | 2016-12-29 | 2017-03-13 | Personal property status assessing method, apparatus, device, and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611246227.XA CN107666649A (en) | 2016-12-29 | 2016-12-29 | Personal property state evaluating method and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107666649A true CN107666649A (en) | 2018-02-06 |
Family
ID=61122475
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201611246227.XA Pending CN107666649A (en) | 2016-12-29 | 2016-12-29 | Personal property state evaluating method and device |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN107666649A (en) |
| WO (1) | WO2018120425A1 (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110781221A (en) * | 2019-09-27 | 2020-02-11 | 同济大学 | Estimation decision support system architecture for concealed property of executed person in court |
| CN110781374A (en) * | 2018-07-13 | 2020-02-11 | 北京字节跳动网络技术有限公司 | User data matching method and device, electronic equipment and computer readable medium |
| CN112383544A (en) * | 2020-11-13 | 2021-02-19 | 西安热工研究院有限公司 | Service behavior portrait-based crawler resisting method suitable for electric power SCADA |
| CN113065739A (en) * | 2021-02-24 | 2021-07-02 | 广州互联网法院 | Executed person fulfillment ability evaluation method and device and electronic equipment |
Families Citing this family (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111049664A (en) * | 2018-10-11 | 2020-04-21 | 中兴通讯股份有限公司 | Network alarm processing method, device and storage medium |
| CN109858957A (en) * | 2019-01-17 | 2019-06-07 | 深圳壹账通智能科技有限公司 | Finance product recommended method, device, computer equipment and storage medium |
| CN110442761A (en) * | 2019-06-21 | 2019-11-12 | 深圳中琛源科技股份有限公司 | A kind of user draws a portrait construction method, electronic equipment and storage medium |
| CN111506733B (en) * | 2020-05-29 | 2022-06-28 | 广东太平洋互联网信息服务有限公司 | Object portrait generation method and device, computer equipment and storage medium |
| CN112925982B (en) * | 2021-03-12 | 2023-04-07 | 上海意略明数字科技股份有限公司 | User redirection method and device, storage medium and computer equipment |
| CN117271905B (en) * | 2023-11-21 | 2024-02-09 | 杭州小策科技有限公司 | Crowd image-based lateral demand analysis method and system |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| 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 |
| CN104965913A (en) * | 2015-07-03 | 2015-10-07 | 重庆邮电大学 | GPS (global positioning system) geographic position data mining based user classification method |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105894089A (en) * | 2016-04-21 | 2016-08-24 | 百度在线网络技术（北京）有限公司 | Method of establishing credit investigation model, credit investigation determination method and the corresponding apparatus thereof |
-
2016
- 2016-12-29 CN CN201611246227.XA patent/CN107666649A/en active Pending
-
2017
- 2017-03-13 WO PCT/CN2017/076463 patent/WO2018120425A1/en active Application Filing
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| 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 |
| CN104965913A (en) * | 2015-07-03 | 2015-10-07 | 重庆邮电大学 | GPS (global positioning system) geographic position data mining based user classification method |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110781374A (en) * | 2018-07-13 | 2020-02-11 | 北京字节跳动网络技术有限公司 | User data matching method and device, electronic equipment and computer readable medium |
| CN110781221A (en) * | 2019-09-27 | 2020-02-11 | 同济大学 | Estimation decision support system architecture for concealed property of executed person in court |
| CN112383544A (en) * | 2020-11-13 | 2021-02-19 | 西安热工研究院有限公司 | Service behavior portrait-based crawler resisting method suitable for electric power SCADA |
| CN112383544B (en) * | 2020-11-13 | 2023-03-24 | 西安热工研究院有限公司 | Service behavior portrait-based crawler resisting method suitable for electric power SCADA |
| CN113065739A (en) * | 2021-02-24 | 2021-07-02 | 广州互联网法院 | Executed person fulfillment ability evaluation method and device and electronic equipment |
| CN113065739B (en) * | 2021-02-24 | 2023-07-04 | 广州互联网法院 | Method and device for evaluating performance capability of executed person and electronic equipment |
Also Published As
| Publication number | Publication date |
|---|---|
| WO2018120425A1 (en) | 2018-07-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107666649A (en) | Personal property state evaluating method and device | |
| Yue et al. | Measurements of POI-based mixed use and their relationships with neighbourhood vibrancy | |
| Fisher et al. | Modelling the errors in areal interpolation between zonal systems by Monte Carlo simulation | |
| Lichman et al. | Modeling human location data with mixtures of kernel densities | |
| CN109949063B (en) | Address determination method and device, electronic equipment and readable storage medium | |
| CN106651603A (en) | Risk evaluation method and apparatus based on position service | |
| Fu et al. | Identifying spatiotemporal urban activities through linguistic signatures | |
| CN112861972B (en) | Site selection method and device for exhibition area, computer equipment and medium | |
| CN105894089A (en) | Method of establishing credit investigation model, credit investigation determination method and the corresponding apparatus thereof | |
| CN107239967A (en) | House property information processing method, device, computer equipment and storage medium | |
| Bokányi et al. | Race, religion and the city: twitter word frequency patterns reveal dominant demographic dimensions in the United States | |
| CN112069281B (en) | Method and device for determining position of interest point, computer equipment and storage medium | |
| CN115759640B (en) | Public service information processing system and method for smart city | |
| Tomal | Analysing the coupling coordination degree of socio-economic-infrastructural development and its obstacles: The case study of Polish rural municipalities | |
| Zandbergen | Influence of street reference data on geocoding quality | |
| Jayasekare et al. | The price of a view: Estimating the impact of view on house prices | |
| Liu et al. | Analysis of the performance and robustness of methods to detect base locations of individuals with geo-tagged social media data | |
| CN107016042B (en) | Address information verification system based on user position log | |
| Roig-Tierno et al. | Business opportunities analysis using GIS: the retail distribution sector | |
| CN111192153A (en) | Crowd relation network construction method and device, computer equipment and storage medium | |
| Rodrigues et al. | Automatic classification of points-of-interest for land-use analysis | |
| Moradi et al. | Exploring five indicators for the quality of OpenStreetMap road networks: A case study of Québec, Canada | |
| Manley et al. | New forms of data for understanding urban activity in developing countries | |
| Groff | Measuring a place’s exposure to facilities using geoprocessing models: An illustration using drinking places and crime | |
| Yuan et al. | Modeling activity spaces using big geo‐data: Progress and challenges |
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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180206 |