CN110413901A - A kind of assessing credit risks method based on social network analysis - Google Patents
A kind of assessing credit risks method based on social network analysis Download PDFInfo
- Publication number
- CN110413901A CN110413901A CN201910713787.9A CN201910713787A CN110413901A CN 110413901 A CN110413901 A CN 110413901A CN 201910713787 A CN201910713787 A CN 201910713787A CN 110413901 A CN110413901 A CN 110413901A
- Authority
- CN
- China
- Prior art keywords
- network
- application part
- social
- social networks
- data
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9536—Search customisation based on social or collaborative filtering
-
- 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
- G06Q40/03—Credit; Loans; Processing thereof
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/01—Social networking
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Databases & Information Systems (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Accounting & Taxation (AREA)
- Finance (AREA)
- General Business, Economics & Management (AREA)
- Strategic Management (AREA)
- Economics (AREA)
- Marketing (AREA)
- Human Resources & Organizations (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Health & Medical Sciences (AREA)
- Tourism & Hospitality (AREA)
- General Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Computing Systems (AREA)
- Development Economics (AREA)
- Technology Law (AREA)
- Financial Or Insurance-Related Operations Such As Payment And Settlement (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
A kind of assessing credit risks method based on social network analysis, it is related to credit financing air control assessment system technical field, the credit financing methods of risk assessment for solving existing credit financing air control assessment system manually relies on program height, in the presence of the low technical deficiency of fraud discrimination, include: data collection and cleaning;Social networks building;Social networks optimization;It is being made into different size of social networks based on history application part, the social networks of preset value is less than for application part, is cut;The building of network Rating Model;Realize the risk quantification to application part;Network graphic is shown.Risk assessment is carried out from new mode, is not only the supplement to current risk appraisal procedure, and have more effective identification for clique or group's fraud.
Description
Technical field
The present invention relates to credit financing air control assessment system technical fields, and in particular to assessing credit risks system
Assessing credit risks method improves aspect.
Background technique
In recent years, with the fast development of credit financing business, while financial general favour masses, also because overdue and promise breaking is given
Financial circles bring massive losses.Especially under the dual-impingement that macroeconomy downlink and supervision are severely administered, bank, China
Industry has fallen into bad soaring predicament, and huge overdue loan burden causes credit resource to be difficult to discharge.Risk management ability
Power will become the following core distinguished bank and manage superiority and inferiority.
Current credit financing methods of risk assessment is mainly based on expertise and traditional machine learning algorithm.Expert
Business personnel is empirically based on to the familiarity of fraud scene, simple in rule, discrimination is not high.Machine learning utilizes great amount of samples
Model training is carried out, fraud discrimination can be improved.But these two kinds of methods all carry out wind just with application part correlated characteristic variable
Control identification does not excavate application part and applies for personage's social network relationships of part behind to carry out risk assessment.
Summary of the invention
In conclusion it is an object of the invention to solve the credit financing risk of existing credit financing air control assessment system
Appraisal procedure manually relies on program height, there is the low technical deficiency of fraud discrimination, and proposes a kind of based on social network analysis
Assessing credit risks method.
In order to solve technical problem proposed by the invention, the technical solution of use are as follows:
A kind of assessing credit risks method based on social network analysis, it is characterised in that the method includes having:
Data collection and cleaning;It collects application part and corresponds to the related social attribute feature of applicant, and cleaned, standard
Change and packet transaction;
Social networks building;To apply for part as center node, related social attribute characteristic information is connecting node, according to not
With application part in the matching relationship of identical connecting node, network building is carried out;
Social networks optimization;It is being made into different size of social networks based on history application part, for application part less than pre-
If the social networks of value, is cut;
The building of network Rating Model;After completing social networks optimization, need to carry out risk score, each application to network
Part will all belong to a network;It is then based on network structure and applies for the correlated characteristic of part, the machining feature factor constructs risk
Rating Model realizes the risk quantification to application part;
Network graphic is shown;Diagram data is generated according to the social network relationships of application part, and is illustrated as social network diagram, into
One step customizes according to demand shows style.
Echnical solution as defined further in the present invention includes:
In the data-gathering process, the related social attribute feature that the application part of collection corresponds to applicant includes:
Based on the related structured and unstructured data of history application part, including address, contact method, circle of friends, contact person, identity
Card, referrer apply for the overdue of part and promise breaking relevant information, the credit information of applicant, black and white lists information.
Data cleansing includes following treatment process:
1. shortage of data and abnormal processing;
2. entity extracts, entity object is extracted from structural data and unstructured data, entity attribute, between entity
Relationship;
3. entity fusion, disappear qi: guaranteeing consistency, merged, disambiguated to unstructured data, standardize and obscure
Match, is then grouped processing, same group of text variable distributes identical group number, and is grouped volume to all text features
Code.
Complete social networks optimization after, the Rating Model to network carry out risk score, using logistic regression into
Application part is labeled as positive negative sample, and according to network struction according to the markup information of history application part by the building of row risk model
Correlated characteristic, training Logic Regression Models are as follows:
Logit (p)=β0+β1x1+β2x2+…+βkxk
The wherein probability of p=P (y=1 | X=x).
The invention has the benefit that the present invention corresponds to the related social attribute feature of applicant by collecting application part,
It then, can be according to application part after social networks building, social networks optimization and network Rating Model building relevant treatment
Social network relationships generate diagram data, and are illustrated as social network diagram, facilitate business personnel according to social network relationships carry out into
The investigation of one step, the present invention carry out risk assessment from new mode, are not only the supplement to current risk appraisal procedure, and
There is more effective identification for clique or group's fraud.
Detailed description of the invention
Fig. 1 is workflow schematic diagram of the invention;
Fig. 2 is social networks map;
Fig. 3 is to generate diagram data, and the social networks map being illustrated as according to the social network relationships of application part.
Specific embodiment
The present invention is further described below in conjunction with attached drawing and currently preferred specific embodiment.
Referring to figs. 1 to shown in Fig. 3, the present invention is based on the assessing credit risks method of social network analysis, feature exists
In the method includes having:
Data collection and cleaning;It collects application part and corresponds to the related social attribute feature of applicant, and cleaned, standard
Change and packet transaction;
Social networks building;To apply for part as center node, related social attribute characteristic information is connecting node, according to not
With application part in the matching relationship of identical connecting node, network building is carried out;Based on this, the social network of history application part is constructed
Network relationship, it will big and small social networks map is formed, shown in Fig. 2;The social network relationships figure will show different Shens
It please connection relationship between part.There is certain association each other in the application part of consolidated network.
Social networks optimization;It is being made into different size of social networks based on history application part, for application part less than pre-
If the social networks of value, is cut;Only one possible application part of some social networks, does not construct with other application parts and joins
System.There may be hundreds and thousands of application parts for some networks, and the connection of network actually exists connection power.In Weak link
Node, need to cut network, form relatively reasonable network group;
The building of network Rating Model;After completing social networks optimization, need to carry out risk score, each application to network
Part will all belong to a network;It is then based on network structure and applies for the correlated characteristic of part, the machining feature factor constructs risk
Rating Model realizes the risk quantification to application part;
Network graphic is shown;Diagram data is generated according to the social network relationships of application part, and is illustrated as social network diagram, into
One step customizes according to demand shows style.It is further investigated in order to facilitate business personnel according to social network relationships, it can
To generate diagram data according to the social network relationships of application part, and it is illustrated as social network diagram.Exhibition is further customized according to demand
Show that style is as shown in Figure 3.
Technical solution of the present invention is described in detail as follows:
1, data collection: social networks related data is compiled.Related structured and non-knot based on history application part
Structure data.Including but not limited to: all kinds of addresses, contact method, circle of friends, contact person, identity card, referrer apply for part
Overdue and promise breaking relevant information, the credit information of applicant, black and white lists information etc..
2, data cleansing and pretreatment.Mainly include following treatment process:
1. shortage of data and abnormal processing.
2. entity extracts.Entity object is extracted from structural data and unstructured data, such as: customer name, customer account
Number;Entity attribute, such as: company name, cell-phone number, home address;Relationship between entity, such as: Peer Relationships, neighborhood, visitor
Family is with holding relationship etc. between cell-phone number.
3. entity fusion, disappear qi: guaranteeing consistency.Unstructured data is merged, is disambiguated, standardizes and obscures
Match, is then grouped processing, same group of text variable distributes identical group number, and is grouped volume to all text features
Code.
3, social networks constructs.According to the good history request for data of packet transaction, to apply for node, application centered on part
The relevant information of part carries out network struction as connecting node, according to nonoriented edge relationship.Big and small social network will be will form
Network map.
4, social networks optimizes.Also network cutting or community discovery are cried.It is different size of being made into based on history application part
Social networks, there may be hundreds and thousands of application parts for some networks, and the connection of network actually exists connection power.In
The node of Weak link needs to cut network, forms the relatively reasonable network group (community) of size.
5, network Rating Model constructs.After completing social networks optimization, need to carry out risk score to network.It is sharp herein
Risk model building is carried out with logistic regression.According to the markup information of history application part, application part is labeled as positive negative sample.And
According to network struction correlated characteristic, training Logic Regression Models.It is as follows:
Logit (p)=β0+β1x1+β2x3+…+βkxk
The wherein probability of p=P (y=1 | X=x).
It can refer to following dimension derivative feature:
1, feature of risk.Apply for number of packages, blacklist application number in network, gray list application number, network in network in network
In overdue application number, promise breaking application number of packages in network, whether application part history occur the information such as overdue, promise breaking;
2, authentication feature." same identity card difference mobile phone " rule triggering number, " same company name different address " rule
Number is triggered, " same mailbox different identity card " rule triggering number etc.;
3, essential attribute feature.The average application age of consolidated network, income etc.;
4, link information feature.Apply for part and blacklist, gray list, overdue, the connection number of promise breaking application part etc.;
6, network is shown.It is further investigated in order to facilitate business personnel according to social network relationships, Ke Yigen
Diagram data is generated according to the social network relationships of application part, and is illustrated as social network diagram.Further customization shows wind according to demand
Lattice.
Claims (4)
1. a kind of assessing credit risks method based on social network analysis, it is characterised in that the method includes having:
Data collection and cleaning;Collect application part correspond to the related social attribute feature of applicant, and cleaned, standardize and
Packet transaction;
Social networks building;To apply for part as center node, related social attribute characteristic information is connecting node, according to different Shens
Please part identical connecting node matching relationship, carry out network building;
Social networks optimization;It is being made into different size of social networks based on history application part, preset value is less than for application part
Social networks, cut;
The building of network Rating Model;After completing social networks optimization, need to carry out risk score, each application part to network
A network will be belonged to;It is then based on network structure and applies for the correlated characteristic of part, the machining feature factor constructs risk score
Model realizes the risk quantification to application part;
Network graphic is shown;Diagram data is generated according to the social network relationships of application part, and is illustrated as social network diagram, further
Customization shows style according to demand.
2. a kind of assessing credit risks method based on social network analysis according to claim 1, it is characterised in that: institute
In the data-gathering process stated, the related social attribute feature that the application part of collection corresponds to applicant includes: being based on history Shen
Please part related structured and unstructured data, including address, contact method, circle of friends, contact person, identity card, referrer,
Apply for the overdue of part and promise breaking relevant information, the credit information of applicant, black and white lists information.
3. a kind of assessing credit risks method based on social network analysis according to claim 1, it is characterised in that: number
Include following treatment process according to cleaning:
1. shortage of data and abnormal processing;
2. entity extracts, entity object is extracted from structural data and unstructured data, entity attribute, the pass between entity
System;
3. entity fusion, disappear qi: guarantee consistency, unstructured data is merged, is disambiguated, standardization and fuzzy matching,
Then it is grouped processing, same group of text variable distributes identical group number, and is grouped coding to all text features.
4. a kind of assessing credit risks method based on social network analysis according to claim 1, it is characterised in that: In
After completing social networks optimization, the Rating Model carries out risk score to network, carries out risk model using logistic regression
Application part is labeled as positive negative sample according to the markup information of history application part by building, and according to network struction correlated characteristic,
Training Logic Regression Models are as follows:
Logit (p)=β0+β1x1+β2x2+…+βkxk
The wherein probability of p=P (y=1 | X=x).
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910713787.9A CN110413901B (en) | 2019-08-02 | 2019-08-02 | Credit risk assessment method based on social network analysis |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910713787.9A CN110413901B (en) | 2019-08-02 | 2019-08-02 | Credit risk assessment method based on social network analysis |
Publications (2)
Publication Number | Publication Date |
---|---|
CN110413901A true CN110413901A (en) | 2019-11-05 |
CN110413901B CN110413901B (en) | 2023-05-26 |
Family
ID=68365571
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910713787.9A Active CN110413901B (en) | 2019-08-02 | 2019-08-02 | Credit risk assessment method based on social network analysis |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110413901B (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113850663A (en) * | 2021-08-24 | 2021-12-28 | 江苏中交车旺科技有限公司 | Data processing method, system, equipment and medium for new user recommendation |
CN114462414A (en) * | 2022-04-11 | 2022-05-10 | 北京微聚未来科技有限公司 | Personal credit evaluation method and system based on nested named entity recognition technology |
CN116861101A (en) * | 2023-06-30 | 2023-10-10 | 深圳市爱彼利科技有限公司 | Data processing method and device for social matching |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20120226590A1 (en) * | 2011-03-01 | 2012-09-06 | Early Warning Services, Llc | System and method for suspect entity detection and mitigation |
CN107943879A (en) * | 2017-11-14 | 2018-04-20 | 上海维信荟智金融科技有限公司 | Fraud group detection method and system based on social networks |
US20180308160A1 (en) * | 2016-02-01 | 2018-10-25 | Tencent Technology (Shenzhen) Company Limited | Risk assessment method and system |
-
2019
- 2019-08-02 CN CN201910713787.9A patent/CN110413901B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20120226590A1 (en) * | 2011-03-01 | 2012-09-06 | Early Warning Services, Llc | System and method for suspect entity detection and mitigation |
US20180308160A1 (en) * | 2016-02-01 | 2018-10-25 | Tencent Technology (Shenzhen) Company Limited | Risk assessment method and system |
CN107943879A (en) * | 2017-11-14 | 2018-04-20 | 上海维信荟智金融科技有限公司 | Fraud group detection method and system based on social networks |
Non-Patent Citations (1)
Title |
---|
熊正德等: "基于有序logistic模型的互联网金融客户违约风险研究", 《系统工程》 * |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113850663A (en) * | 2021-08-24 | 2021-12-28 | 江苏中交车旺科技有限公司 | Data processing method, system, equipment and medium for new user recommendation |
CN114462414A (en) * | 2022-04-11 | 2022-05-10 | 北京微聚未来科技有限公司 | Personal credit evaluation method and system based on nested named entity recognition technology |
CN116861101A (en) * | 2023-06-30 | 2023-10-10 | 深圳市爱彼利科技有限公司 | Data processing method and device for social matching |
CN116861101B (en) * | 2023-06-30 | 2024-03-29 | 深圳市爱彼利科技有限公司 | Data processing method and device for social matching |
Also Published As
Publication number | Publication date |
---|---|
CN110413901B (en) | 2023-05-26 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109918511B (en) | BFS and LPA based knowledge graph anti-fraud feature extraction method | |
CN110413901A (en) | A kind of assessing credit risks method based on social network analysis | |
CN108492173A (en) | A kind of anti-Fraud Prediction method of credit card based on dual-mode network figure mining algorithm | |
Umayaparvathi et al. | Applications of data mining techniques in telecom churn prediction | |
CN110413707A (en) | The excavation of clique's relationship is cheated in internet and checks method and its system | |
CN110009475A (en) | Risk checks method for monitoring, device, computer equipment and storage medium | |
CN108009926A (en) | Method, information processor and readable storage medium storing program for executing for user's classification | |
CN110837963A (en) | Risk control platform construction method based on data, model and strategy | |
CN103796183B (en) | A kind of refuse messages recognition methods and device | |
CN107368918A (en) | Data processing method and device | |
Bhatnagar et al. | Analytical customer relationship management in insurance industry using data mining: a case study of Indian insurance company | |
CN112330047A (en) | Credit card repayment probability prediction method based on user behavior characteristics | |
CN107977855B (en) | Method and device for managing user information | |
CN112508619A (en) | Internet marketing cloud service platform | |
CN112907356A (en) | Overdue collection method, device and system and computer readable storage medium | |
Iranmanesh et al. | Customer churn prediction using artificial neural network: An analytical CRM application | |
CN110648217A (en) | Wind control system based on big data and artificial intelligence | |
CN117196630A (en) | Transaction risk prediction method, device, terminal equipment and storage medium | |
WO2024001102A1 (en) | Method and apparatus for intelligently identifying family circle in communication industry, and device | |
CN110324418A (en) | Method and apparatus based on customer relationship transmission service | |
CN108564380B (en) | Telecommunication user classification method based on iterative decision tree | |
CN111465021B (en) | Graph-based crank call identification model construction method | |
Xu et al. | Big data analytics toward intelligent mobile service provisions of customer relationship management in e-commerce | |
Li et al. | Influence of Internet-based Social Big Data on Personal Credit Reporting | |
CN113704637A (en) | Object recommendation method, device and storage medium based on artificial intelligence |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |