CN112419074A - Vehicle insurance fraud group identification method and device - Google Patents
Vehicle insurance fraud group identification method and device Download PDFInfo
- Publication number
- CN112419074A CN112419074A CN202011272409.0A CN202011272409A CN112419074A CN 112419074 A CN112419074 A CN 112419074A CN 202011272409 A CN202011272409 A CN 202011272409A CN 112419074 A CN112419074 A CN 112419074A
- Authority
- CN
- China
- Prior art keywords
- risk
- vehicle insurance
- data
- insurance
- vehicle
- 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
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
- G06Q40/08—Insurance
-
- 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
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0635—Risk analysis of enterprise or organisation activities
Abstract
The invention discloses a vehicle insurance fraud gang identification method and a device, wherein the method comprises the steps of obtaining vehicle insurance data, and carrying out visual processing on the vehicle insurance data to obtain map data corresponding to the vehicle insurance data; grouping and dividing the map data through a map community discovery algorithm to obtain a plurality of groups of vehicle insurance grouping data; carrying out visualization processing on the multiple groups of vehicle insurance partnership data to obtain a vehicle insurance partnership map; and performing risk evaluation on the car insurance ganged partner graph through a preset risk case scoring rule, and determining that the car insurance ganged partners possibly being fraud ganged partners in the car insurance ganged partner graph according to a risk evaluation result. The method aims to convert the vehicle insurance data into graph data with a graph structure, divide the gangues through a graph community discovery algorithm, and carry out risk scoring on the vehicle insurance gangues through the constructed risk case rules to quickly obtain the vehicle insurance gangues which are possibly cheated, does not depend on manual experience any more, has high accuracy, greatly reduces the cost, and solves the problems of low vehicle insurance cheating gangue identification efficiency and low accuracy.
Description
Technical Field
The invention relates to the technical field of computers, in particular to a vehicle insurance fraud group identification method and device.
Background
In the internet + age, the number of vehicle insurance group fraud cases is increasing, causing losses of different degrees to investors, companies and countries. Companies and organizations are urgently seeking a technical scheme to find ganged fraud cases, and prevent the ganged fraud cases from being lost and prevent and recover the ganged fraud cases in time.
Most of existing vehicle insurance fraud group identification methods adopt a manual investigation mode to avoid fraud and leakage risks and compress claim settlement water. However, manual investigation has the following disadvantages:
1) the case audit measurement is large, the audit work difficulty is high, the audit efficiency is low, and the cost input and the output are asymmetric;
2) depending on personal experience, the auditing and surveying threshold is high, and the difference between artificial judgment and reality is large, so that misjudgment and more missed judgment phenomena are caused;
3) and the analysis can be based on data surface analysis only, and the depth and dimensionality of the analysis are insufficient.
Based on the above, the accuracy and the efficiency of the finally determined vehicle insurance fraud group are poor.
Accordingly, the prior art is yet to be developed and improved.
Disclosure of Invention
Therefore, it is necessary to provide a vehicle insurance fraud group identification method and device aiming at the technical problems of low efficiency and poor accuracy of the existing vehicle insurance fraud group identification method.
A vehicle insurance fraud group identification method, the method comprising:
acquiring vehicle insurance data, and processing the vehicle insurance data to obtain map data corresponding to the vehicle insurance data;
grouping and dividing the map data through a map community discovery algorithm to obtain a plurality of groups of vehicle insurance grouping data;
carrying out visualization processing on the multiple groups of vehicle insurance partnership data to obtain a vehicle insurance partnership map;
and performing risk evaluation on the car insurance ganged partner graph through a preset risk case scoring rule, and determining that the car insurance ganged partners possibly being fraud ganged partners in the car insurance ganged partner graph according to a risk evaluation result.
The car insurance fraud gang identification method includes the following steps of:
acquiring initial vehicle insurance data;
preprocessing the initial vehicle insurance data to obtain vehicle insurance data;
and importing the vehicle insurance data into a visual data exchange platform so as to convert the vehicle insurance data into graph data conforming to a graph database storage structure, wherein the graph data at least comprises a report number, an insured person and a target drive, and the report number is used for distinguishing vehicle insurance groups.
The car insurance fraud group identification method is characterized in that the map community discovery algorithm at least comprises one of a group search algorithm and a connection component algorithm.
The vehicle insurance fraud group partner identification method comprises a plurality of nodes and connection relations among the nodes, wherein risk roles of the nodes comprise one or more of a report telephone, a target person, a target vehicle, three vehicles, three persons, a wounded person, a insured person and a person, and the connection relations comprise one or more of the insured person, the three vehicles, the three driving and the target driving.
The car insurance fraud group identification method comprises the steps that preset risk case scoring rules comprise judgment rules and operation rules, the judgment rules comprise risk types, risk categories, risk category grades and risk category grades, each risk type corresponds to a plurality of risk categories, each risk category corresponds to a plurality of risk category grades, the risk types at least comprise cross risks, risk risks of the three, target risk and other factor risks, and the operation rules comprise risk category basic scores and risk category grades and risk category cardinality.
The vehicle insurance fraud group identification method comprises the following steps of carrying out risk evaluation on the vehicle insurance group map through a preset risk case scoring rule, and determining vehicle insurance groups which may be fraud groups in the vehicle insurance group map according to a risk evaluation result:
acquiring each node of the same vehicle insurance group in the vehicle insurance group map and the role of each node;
determining the risk major categories of the nodes according to the roles of the nodes;
if the same node belongs to a plurality of risk categories, determining a risk category basic classification corresponding to each risk category of the node;
selecting a risk major corresponding to the maximum value in the determined risk major basis scores as a target risk major of the node;
counting the risk occurrence times of target roles corresponding to the target risk categories corresponding to the nodes;
determining a target risk major grade and a target risk major grade corresponding to each node according to a preset judgment rule and the risk occurrence frequency of each node;
risk scoring is carried out on each node through a preset operation rule so as to obtain risk score of each node;
determining target risk scores of the same vehicle insurance ganged partner according to the risk scores of all nodes in the same vehicle insurance ganged partner;
and determining the vehicle insurance groups which are possibly cheating groups according to the target risk scores of the vehicle insurance groups.
The vehicle insurance fraud group identification method comprises the following steps of:
comparing the target risk score of each vehicle insurance group with a preset risk early warning threshold value;
and if the target risk score exceeds a preset risk early warning threshold value, determining that the vehicle insurance group corresponding to the target risk score is possibly a fraud vehicle insurance group, sending suspicious case data and case risk analysis reports corresponding to the vehicle insurance group to an expert end, and informing corresponding offline expert investigation services to qualitatively determine case risks.
The vehicle insurance fraud group identification method comprises the following steps of carrying out risk evaluation on the vehicle insurance group map through preset risk case scoring rules, and determining that the vehicle insurance group map is possibly a fraud group according to a risk evaluation result:
obtaining a vehicle insurance database, wherein the vehicle insurance database at least comprises a high risk component database and a blacklist database;
matching the high-risk component library according to the risk scoring result of each vehicle insurance group to determine whether high-risk cases exist in the vehicle insurance group;
and matching the blacklist database according to the risk scoring result of each vehicle insurance group to determine the vehicle insurance groups possibly in the blacklist database.
The car insurance fraud group identification method comprises the following steps:
and predicting the graph data by utilizing a pre-trained evaluation model, and outputting each vehicle insurance group, whether each vehicle insurance group is cheated or not and the cheating probability through the evaluation model.
The application also provides a car insurance fraud group identification device, which comprises a memory and a processor, wherein the memory stores a computer program, and the computer program causes the processor to execute the steps in the car insurance fraud group identification method when being executed by the processor.
The embodiment of the invention has the following beneficial effects:
the invention discloses a vehicle insurance fraud group partner identification method and device, which are used for converting vehicle insurance data into graph data with a graph structure, dividing group partners through a graph community discovery algorithm, and carrying out risk scoring on the vehicle insurance group partners through a constructed risk case rule, so as to quickly obtain vehicle insurance group partners which are possibly fraudulent, do not depend on manual experience any more, have high accuracy, greatly reduce the cost and solve the problems of low vehicle insurance fraud group partner identification efficiency and low accuracy.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.
Wherein:
fig. 1 is a flowchart of an embodiment of a car insurance fraud group identification method provided by the present invention.
Fig. 2 is a partial schematic view of an insurance ganged partner identifying method in the invention.
Fig. 3 is a block diagram of a car insurance fraud group identification apparatus provided by the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Referring to fig. 1, fig. 1 illustrates a flowchart of an embodiment of a car insurance fraud group identification method provided by the present invention. As shown in fig. 1, the car insurance fraud group identification method includes:
and S10, acquiring the vehicle insurance data, and processing the vehicle insurance data to obtain map data corresponding to the vehicle insurance data.
Specifically, the vehicle insurance data refers to various data involved in the occurrence of a vehicle accident. The vehicle insurance data can be obtained through various channels, such as electronic application data mentioned by a user when related vehicle loan is carried out, submitted handwritten application data or related data obtained by crawling on the internet. The form of acquiring the vehicle insurance data is not particularly limited in the present embodiment.
The standard format for the vehicle insurance data is typically.xls. The car insurance data comprises a branch company, a case number, a case reporter, a case call, a insured person identification number, a target driver identification number, a target license number, three drivers, an identification number, three license numbers, a wounded name, a wounded identification number, a target car type, a target frame number, a target driving license main mark identification number, a target repair plant, a target damage amount, a target damage person, three frame numbers, three car repair plant
The three loss settlement amounts, the three loss settlement members, the three underwriting companies settled amount, the surveyor, the bank for making an account with the name of the lost account, the bank account number, the insurance start period, the insurance end period, the report time, the insurance time, the post-sale mark, whether to press or not, and the like. It should be noted that the vehicle insurance data must include a notice number, an insured life, and a target driving. The case number is used for determining case occurrence and distinguishing vehicle insurance groups. Namely, the related case-related information of the vehicle accident can be known through the case number.
Since the acquired vehicle insurance data may have various contents, the vehicle insurance data needs to be preprocessed to reduce invalid operations and improve accuracy. The preprocessing comprises check field format, format conversion, error correction such as null removal processing, duplicate removal processing, decimal processing and the like. For example: the process of going to the air refers to screening out key fields, namely, a report number, a insured person and vehicle insurance data with the target driving as a null value for prompting or deleting.
Further, the acquiring the vehicle insurance data and processing the vehicle insurance data to obtain map data corresponding to the vehicle insurance data specifically includes:
s11, acquiring initial vehicle insurance data;
s12, preprocessing the initial vehicle insurance data to obtain vehicle insurance data;
s13, importing the vehicle insurance data into a visual data exchange platform so that the vehicle insurance data are converted into graph data conforming to a graph database storage structure.
Specifically, the visualization data exchange platform employs a visualization tool, ETL, for processing. And converting the vehicle insurance data into graph data conforming to a graph database storage structure. The graph database stores nodes corresponding to graph data and connection relations between the nodes.
In this embodiment, the nodes are used to represent users, and different users are distinguished by different roles. Thus, the roles of the nodes include one or more of a notice telephone, a target person, a target vehicle, a three-person, an injured person, an insured person and a person, and the connection relationship includes one or more of an insured person, a three-person vehicle, a three-person drive and a target drive.
The connection relationship of the specific nodes (the expression form is n-r- > m, wherein n and m are nodes, and r is the connection relationship) is as follows:
report telephone-insured person
Three-in-three vehicle for reporting case and telephone
Reporting the case, calling the person, and being injured
Reporting telephone-three driving- > man
Reporting telephone-target driving-person
Target person-target drive- > target vehicle
Three person-three driving-three vehicle
According to the nodes and the connection relation among the nodes, the car insurance graph corresponding to the car insurance data can be displayed. That is, the drawing data is imported into the map library and displayed as a car insurance map. The car insurance graph comprises a plurality of nodes, and edges among the nodes are connection relations.
S20, carrying out group division on the map data through a map community discovery algorithm to obtain multiple groups of vehicle insurance group data.
In particular, the graph community discovery algorithm is a class of efficient data structures that solve the dynamic connectivity problem. In this embodiment, the map community discovery algorithm employs a union set search algorithm, a connection component algorithm, and the like. In computer science, a union query is a tree-shaped data structure used for processing a non-intersection combination and query problem, and has two operations (1, union; 2, find search) and a problem needing to be solved (1, whether isconnected is connected with each other or not, and whether isSameSet is in the same set or not). Therefore, the group division can be carried out on the graph data through the combined search and set search algorithm, so that at least one group of vehicle insurance group data is obtained. Therefore, labor cost is reduced, searching time is shortened, and dividing efficiency is improved. It should be noted that the map community discovery algorithm is not limited, as long as the two operations and the algorithm of a question to be solved are satisfied.
Further, in order to facilitate the subsequent fraud group identification, the minimum unit of the vehicle insurance group, namely the minimum threshold node number, is set, and the minimum unit refers to the possibility of meeting the fraud group. Therefore, the number of the nodes of the multiple groups of vehicle insurance partnership data obtained by dividing the vehicle insurance partnership is not larger than the preset threshold value, so that a large number of nodes do not exist in the divided vehicle insurance partnership data, and the subsequent cheating partnership evaluation determines that the cheating partnership is more accurate. And the data processing amount of the vehicle insurance group data divided into the minimum units is reduced, and the dividing efficiency is accelerated.
S30, carrying out visualization processing on the multiple groups of car insurance ganged data to obtain a car insurance ganged map.
Specifically, the visualization tool ETL is used to perform visualization processing on the multiple sets of car insurance partnership data obtained in step S20, so as to obtain a car insurance partnership map. The car insurance partnership map includes a plurality of sets of car insurance partnership data. Similarly, the car insurance group map comprises a plurality of nodes and connection relations among the nodes. The car insurance partnership project is displayed on a display screen to visually determine which nodes belong to the same car insurance partnership as shown in figure 2.
S40, carrying out risk evaluation on the car insurance ganged partner graph through preset risk case scoring rules, and determining that the car insurance ganged partner graph is possibly a fraud ganged partner according to the risk evaluation result.
Specifically, the preset risk case scoring rule is a rule summarized by expert experience and obtained by analyzing actual car insurance groups. The preset risk case scoring rule can be set in a self-defined mode. The custom rules form a custom rule base. The method comprises the steps of receiving user-defined rule information uploaded by a user through a user side (such as a computer smart phone) and establishing a user-defined rule engine base corresponding to a user-defined rule model according to the rule information. Aiming at the conditions of different users and different regional environments, thousands of risk factors related to each flow link of the car insurance claims can be collected to establish a risk factor library, and the risk factor library can be utilized by the users, and the risk factor library is customized to accord with the risk rule library of the users by freely combining different factors and parameters and is applied to the risk rating of each link of the car insurance claims. The risk factors comprise name telephone information of the party, vehicle model, insurance leaving record, license plate number, reporting time, insurance leaving time and the like. For example, a new custom rules engine library is first constructed; selecting claim settlement links and factors, inputting corresponding parameters, and setting custom rules.
In this embodiment, the preset risk case scoring rule includes a determination rule and an operation rule, the determination rule includes a risk type, a risk category level, and a risk category level, each risk type corresponds to a plurality of risk categories, each risk category corresponds to a plurality of risk category levels, and the risk type at least includes a cross risk, an risk of the three, a target risk, and other risk factors. The operation rules are risk major base classification + risk major grade level. In this embodiment, the risk major base number is set to be 5, which is not limited, and the risk major base number may be set by user according to requirements, such as setting the risk major base number to be 6, 7, and the like.
As shown in table 1 below, table 1 illustrates the risk case scoring rules corresponding to the four risk types.
TABLE 1
For example, the performing risk evaluation on the car insurance partnership project through preset risk case scoring rules, and determining that a car insurance partnership project which may be a fraud partnership in the car insurance partnership project according to a risk evaluation result specifically includes:
s41, acquiring each node of the same vehicle insurance group in the vehicle insurance group map and the role of each node;
s42, determining the risk categories of the nodes according to the roles of the nodes;
s43, if the risk categories to which the same node belongs are multiple, determining the risk category basic classification corresponding to each risk category of the node;
s44, selecting the risk major corresponding to the maximum value in the determined risk major basis scores as the target risk major of the node;
s45, counting the risk occurrence times of the target roles corresponding to the target risk categories corresponding to the nodes;
s46, determining a target risk major grade and a target risk major grade corresponding to each node according to a preset judgment rule and the risk occurrence frequency of each node;
s47, performing risk scoring on each node through a preset operation rule to obtain a risk score of each node;
s48, determining the target risk score of the same vehicle insurance group according to the risk scores of all nodes in the same vehicle insurance group;
s49, determining the vehicle insurance partners that are likely to be fraudulent based on the target risk scores for the respective vehicle insurance partners.
Specifically, assume that a group a includes node 1, node 2, and node 3, and that a vehicle accident occurs such that node 2 hits node 1 and node 1 hits node 3. And determining that the role of the node 1 is the driver of the three nodes and the target driver. Thus, the risk score is calculated taking node 1 as an example. As can be seen from table look-up 1, the target driver with the risk category as the target in the cross risk category is used as the tri-driver and the tri-driver is used as the target driver, and since the basic score 69 of the target driver as the tri-driver is greater than the basic score 68 of the tri-driver as the target driver, the target driver with the risk category as the target in node 1 is determined as the tri-driver. Then, the number of times of taking out the risk of the driver of the node 1 as the three is counted. If the number of times of risk of the node 1 is counted to be 2, according to the operation rule, the risk score of the node 1 is the basic score of the target risk major class + the base number of the target risk major class grade is 69+1 + 5-74.
As can be seen from table 1, the risk category corresponding to the node 1 in the three risk categories is the number of times that the driver of the vehicle is in danger, and for this category, the risk score of the node 1 is calculated as 64+1 × 5 — 69 in the same manner.
As can be seen from table 1, the risk category corresponding to node 1 in the target risk category is the target number of times of risk of the driver, and for this category, the risk score of node 1 is calculated as 59+1 × 5 as 64.
Therefore, node 1 calculates the risk scores 74,69, and 64, and then takes the maximum value of the risk scores as the final risk score of node 1, i.e., the final risk score of node 1 is 74.
And then calculating risk scores corresponding to the roles of other nodes in the same vehicle insurance partnership based on the same principle of the node 1, and weighting and summing the risk scores according to preset role weights to obtain a target risk score of the same vehicle insurance partnership.
Of course, in some embodiments, the maximum or average value of the risk scores in the same insurance partnership is selected as the target risk score for that insurance partnership.
Then, the vehicle insurance gangues which are possibly cheating gangues are determined according to the target risk scores of the vehicle insurance gangues. Specifically, target risk scores of each car insurance ganged case are ranked to obtain ranking results of each car insurance ganged case; and taking the vehicle insurance gangs corresponding to the first N vehicle insurance gangs in the sequencing result as fraud gangs, wherein the initial value of N is 1, and N is a positive integer. In this embodiment, the target risk scores are ranked from high to low, and the top N car insurance partners are selected as fraud partners. The ranking rule is not limited, for example, ranking from low to high, and selecting the last N car insurance teams as fraud teams.
In some embodiments, a potentially fraudulent party car insurance team may be determined by a preset risk pre-warning threshold. The method specifically comprises the following steps: comparing the target risk score of each vehicle insurance group with a preset risk early warning threshold value; and if the target risk score exceeds a preset risk early warning threshold value, determining that the vehicle insurance ganged partner corresponding to the target risk score is possibly a fraud vehicle insurance ganged partner.
And then, after determining the possibly fraudulent car insurance gangs, sending the case data and the fraud analysis report corresponding to the possibly fraudulent car insurance gangs to an expert end, notifying the corresponding offline expert to investigate and service, and carrying out qualitative case risk.
Or matching case data corresponding to the possibly fraudulent car insurance groups with a preset blacklist database to screen out the possibly fraudulent car insurance groups in the blacklist database and generate a corresponding analysis report.
In addition, in some embodiments, the evaluation model is trained by using a machine learning method, so that the car insurance data is imported into the evaluation model, and the graph data is predicted, and car insurance groups, whether the car insurance groups are cheated or not and the cheating probability are output through the evaluation model.
In the present embodiment, fraud prediction is performed by an Xgboost (extreme gradient boost) algorithm. The training method of the evaluation model specifically comprises the following steps:
constructing a database: and establishing a database according to preset maintenance factory data, vehicle owner blacklist data, common survey case data, case-involved telephone data and concluded fraud case data. The database includes a hand of site survey data from different insurance company customers accumulated over the years, case data that has been characterized as fraudulent, and the like.
And (3) constructing a risk rule base, namely establishing a risk rule engine base comprising a plurality of corresponding risk rule models by adopting a data modeling technology according to preset risk rules, training the risk rule models according to data in the database, and modifying the corresponding risk rule engine base according to a training result. For example, a Newton-Raphfilson iterative algorithm and an L-BFGS algorithm can be used as iterative algorithms of a risk rule model to train a big data model of the risk rule; after the iteration of the learning algorithm is completed, the weight corresponding to each attribute of the risk rule can be obtained, then the significance between the existing attribute and the response variable is checked, the attribute set corresponding to the existing training model is verified, and the feature that the significance does not accord with the preset threshold value is deleted. For example, the risk rule engine base comprises more than one hundred of report receiving risk rule bases, document risk rule bases, investigation risk rule bases, damage assessment risk rule engines, audit risk rule engines and the like, and covers the whole process of vehicle insurance claim settlement. The risk elements covered by embodiments of the present invention include the fields listed in step S10.
And constructing an initial model, training the initial model through a training set in a database, and acquiring an evaluation model meeting the standard based on a risk rule base and a loss function.
The specific application scenario is as follows:
vue is adopted at the front end of the whole recognition system to construct a progressive framework of a user interface, and a springboot website is adopted at the rear end of the view to develop and integrate neo4j, phonix, hbase, hdfs, spark and other big data technologies to efficiently process huge historical case data and real-time case data generated by a service system. And the data analysis and data visualization technology is adopted to assist business personnel in analyzing case data and group data. Business personnel only need to import data on a page, after data processing is completed, according to the report, the statistical data of all dimensions, such as dimension information of high-risk cases (high score), high-risk group partners, triggering blacklists, high-risk accessories and the like, are analyzed and checked to check the data of the whole case, the generated car insurance group figures are checked, the condition of the whole group partners is checked, and case clues are searched.
Thus, based on the steps S10-S40, the vehicle insurance group risk identification method aims to convert vehicle insurance data into graph data with a graph structure, divide groups through a graph community discovery algorithm, and carry out risk scoring on the vehicle insurance groups through the constructed risk case rules, so that vehicle insurance groups which are possibly cheat are obtained quickly, manual experience is not relied on, the accuracy is high, the cost is greatly reduced, and the problems of low vehicle insurance cheating group identification efficiency and low accuracy are solved.
In the above embodiments, all or part of the implementation may be realized by software, hardware, firmware, or any combination thereof. When implemented using a software program, may be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When loaded and executed on a computer, cause the processes or functions described in accordance with the embodiments of the invention to occur, in whole or in part. The computer may be a general purpose computer, a special purpose computer, a network of computers, or other programmable device. The computer instructions may be stored in a computer readable storage medium or transmitted from one computer readable storage medium to another, for example, the computer instructions may be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, Digital Subscriber Line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device, such as a server, a data center, etc., that incorporates one or more of the available media. The usable medium may be a magnetic medium (e.g., floppy Disk, hard Disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., Solid State Disk (SSD)), among others.
Based on the method, the application also provides a vehicle insurance fraud group identification device. In one embodiment, as shown in fig. 3, fig. 3 shows a block diagram of a car insurance fraud group recognition apparatus. Which includes at least one processor (processor) 20; a display screen 21; and a memory (memory)22, and may further include a communication Interface (Communications Interface)23 and a bus 24. The processor 20, the display 21, the memory 22 and the communication interface 23 can communicate with each other through the bus 24. The display screen 21 is configured to display a user guidance interface preset in the initial setting mode. The communication interface 23 may transmit information. The processor 20 may call logic instructions in the memory 22 to perform the methods in the embodiments described above.
Furthermore, the logic instructions in the memory 22 may be implemented in software functional units and stored in a computer readable storage medium when sold or used as a stand-alone product.
The memory 22, which is a computer-readable storage medium, may be configured to store a software program, a computer-executable program, such as program instructions or modules corresponding to the methods in the embodiments of the present invention. The processor 20 executes the functional application and data processing, i.e. implements the method in the above-described embodiments, by executing the software program, instructions or modules stored in the memory 22.
The memory 22 may include a storage program area and a storage data area, wherein the storage program area may store an application program required for operating the voice customer service system, at least one function; the storage data area may store data created according to the use of the terminal device, and the like. Further, the memory 22 may include a high speed random access memory and may also include a non-volatile memory. For example, a variety of media that can store program codes, such as a usb disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk, or an optical disk, may also be transient storage media.
The above disclosure is only for the purpose of illustrating the preferred embodiments of the present invention, and it is therefore to be understood that the invention is not limited by the scope of the appended claims.
Claims (10)
1. A vehicle insurance fraud group identification method, the method comprising:
acquiring vehicle insurance data, and processing the vehicle insurance data to obtain map data corresponding to the vehicle insurance data;
grouping and dividing the map data through a map community discovery algorithm to obtain a plurality of groups of vehicle insurance grouping data;
carrying out visualization processing on the multiple groups of vehicle insurance partnership data to obtain a vehicle insurance partnership map;
and performing risk evaluation on the car insurance ganged partner graph through a preset risk case scoring rule, and determining that the car insurance ganged partners possibly being fraud ganged partners in the car insurance ganged partner graph according to a risk evaluation result.
2. The vehicle insurance fraud group identification method according to claim 1, wherein the obtaining vehicle insurance data and processing the vehicle insurance data to obtain map data corresponding to the vehicle insurance data specifically comprises:
acquiring initial vehicle insurance data;
preprocessing the initial vehicle insurance data to obtain vehicle insurance data;
and importing the vehicle insurance data into a visual data exchange platform so as to convert the vehicle insurance data into graph data conforming to a graph database storage structure.
3. The vehicle insurance fraud group identification method according to claim 1, wherein the map community discovery algorithm comprises at least one of a union lookup algorithm and a connection component algorithm.
4. The vehicle insurance fraud group identification method according to claim 1, wherein the vehicle insurance group map comprises a plurality of nodes and connection relations between the nodes, the roles of the nodes comprise one or more of a notice telephone, a target person, a target vehicle, a tri-person, a wounded person, a insured person and a person, and the connection relations comprise one or more of a insured person, a tri-vehicle, a tri-driving and a target driving.
5. The vehicle insurance fraud group identification method according to claim 4, wherein the preset risk case scoring rules include decision rules and calculation rules, the decision rules include risk types, risk categories, risk category grades and risk category grades, each risk type corresponds to a plurality of risk categories, each risk category corresponds to a plurality of risk category grades, the risk types at least include cross risks, three risk risks, target risk and other factor risks, and the calculation rules are risk category basis scores + risk category grades risk category cardinality.
6. The method as claimed in claim 5, wherein the step of performing risk evaluation on the car insurance partnership project according to preset risk case scoring rules and determining the car insurance partnership projects which may be fraudulent parties in the car insurance partnership project according to the risk evaluation result specifically comprises:
acquiring each node of the same vehicle insurance group in the vehicle insurance group map and the role of each node;
determining the risk major categories of the nodes according to the roles of the nodes;
if the same node belongs to a plurality of risk categories, determining a risk category basic classification corresponding to each risk category of the node;
selecting a risk major corresponding to the maximum value in the determined risk major basis scores as a target risk major of the node;
counting the risk occurrence times of target roles corresponding to the target risk categories corresponding to the nodes;
determining a target risk major grade and a target risk major grade corresponding to each node according to a preset judgment rule and the risk occurrence frequency of each node;
risk scoring is carried out on each node through a preset operation rule so as to obtain risk score of each node;
determining target risk scores of the same vehicle insurance ganged partner according to the risk scores of all nodes in the same vehicle insurance ganged partner;
and determining the vehicle insurance groups which are possibly cheating groups according to the target risk scores of the vehicle insurance groups.
7. The vehicle insurance fraud group identification method according to claim 6, wherein the determining vehicle insurance groups that are likely to be fraud groups according to the target risk score of each vehicle insurance group specifically comprises:
ranking the target risk scores of the vehicle insurance ganged cases to obtain a ranking result of each vehicle insurance ganged case;
and taking the vehicle insurance gangs corresponding to the first N vehicle insurance gangs in the sequencing result as fraud gangs, wherein the initial value of N is 1, and N is a positive integer.
8. The method for identifying fraud in car insurance groups as claimed in claim 1, wherein the step of risk evaluating the car insurance group map according to preset risk case scoring rules and determining fraud group in the car insurance group map according to the result of risk evaluation comprises:
case data corresponding to the car insurance gangs which are possibly fraud gangs and fraud analysis reports are sent to an expert end, corresponding off-line expert investigation services are notified, and case risks are qualified;
or matching case data corresponding to the vehicle insurance groups which are possibly fraudulent groups with a preset blacklist database to screen out the vehicle insurance groups which are possibly in the blacklist database and generate a corresponding analysis report.
9. The vehicle insurance fraud group identification method according to claim 1, characterized in that the method further comprises:
and predicting the graph data by utilizing a pre-trained evaluation model, and outputting each vehicle insurance group, whether each vehicle insurance group is cheated or not and the cheating probability through the evaluation model.
10. An insurance fraud group identification apparatus, characterized in that it comprises a memory and a processor, the memory storing a computer program which, when executed by the processor, causes the processor to perform the steps in the insurance fraud group identification method according to any one of claims 1 to 9.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011272409.0A CN112419074A (en) | 2020-11-13 | 2020-11-13 | Vehicle insurance fraud group identification method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011272409.0A CN112419074A (en) | 2020-11-13 | 2020-11-13 | Vehicle insurance fraud group identification method and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112419074A true CN112419074A (en) | 2021-02-26 |
Family
ID=74832081
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011272409.0A Pending CN112419074A (en) | 2020-11-13 | 2020-11-13 | Vehicle insurance fraud group identification method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112419074A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113743954A (en) * | 2021-06-29 | 2021-12-03 | 阳光保险集团股份有限公司 | Vehicle insurance risk network identification method and device, electronic equipment and medium |
| CN114418780A (en) * | 2022-03-11 | 2022-04-29 | 太平金融科技服务(上海)有限公司深圳分公司 | Method, apparatus, computer device and storage medium for identifying fraudulent groups |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107194623A (en) * | 2017-07-20 | 2017-09-22 | 深圳市分期乐网络科技有限公司 | A kind of discovery method and device of clique's fraud |
| CN108334647A (en) * | 2018-04-12 | 2018-07-27 | 阿里巴巴集团控股有限公司 | Data processing method, device, equipment and the server of Insurance Fraud identification |
| CN108734479A (en) * | 2018-04-12 | 2018-11-02 | 阿里巴巴集团控股有限公司 | Data processing method, device, equipment and the server of Insurance Fraud identification |
| CN109165849A (en) * | 2018-08-27 | 2019-01-08 | 众安信息技术服务有限公司 | Methods of risk assessment and device |
| CN110209660A (en) * | 2019-06-10 | 2019-09-06 | 北京阿尔山金融科技有限公司 | Cheat clique's method for digging, device and electronic equipment |
| CN111292008A (en) * | 2020-03-03 | 2020-06-16 | 电子科技大学 | Privacy protection data release risk assessment method based on knowledge graph |
-
2020
- 2020-11-13 CN CN202011272409.0A patent/CN112419074A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107194623A (en) * | 2017-07-20 | 2017-09-22 | 深圳市分期乐网络科技有限公司 | A kind of discovery method and device of clique's fraud |
| CN108334647A (en) * | 2018-04-12 | 2018-07-27 | 阿里巴巴集团控股有限公司 | Data processing method, device, equipment and the server of Insurance Fraud identification |
| CN108734479A (en) * | 2018-04-12 | 2018-11-02 | 阿里巴巴集团控股有限公司 | Data processing method, device, equipment and the server of Insurance Fraud identification |
| CN109165849A (en) * | 2018-08-27 | 2019-01-08 | 众安信息技术服务有限公司 | Methods of risk assessment and device |
| CN110209660A (en) * | 2019-06-10 | 2019-09-06 | 北京阿尔山金融科技有限公司 | Cheat clique's method for digging, device and electronic equipment |
| CN111292008A (en) * | 2020-03-03 | 2020-06-16 | 电子科技大学 | Privacy protection data release risk assessment method based on knowledge graph |
Non-Patent Citations (2)
| Title |
|---|
| 倪思嘉;: "图数据库在商业银行反欺诈审计领域的应用", 上海商业, no. 07 * |
| 周晓楠;黄磊;王飞跃;储明;黄涛;: "图数据库在识别重大疾病保险团伙式欺诈中的应用研究", 保险研究, no. 09 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113743954A (en) * | 2021-06-29 | 2021-12-03 | 阳光保险集团股份有限公司 | Vehicle insurance risk network identification method and device, electronic equipment and medium |
| CN113743954B (en) * | 2021-06-29 | 2024-04-02 | 阳光保险集团股份有限公司 | Vehicle risk network identification method and device, electronic equipment and medium |
| CN114418780A (en) * | 2022-03-11 | 2022-04-29 | 太平金融科技服务(上海)有限公司深圳分公司 | Method, apparatus, computer device and storage medium for identifying fraudulent groups |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110417721B (en) | Security risk assessment method, device, equipment and computer readable storage medium | |
| CN106779457A (en) | A kind of rating business credit method and system | |
| CN109345417B (en) | Online assessment method and terminal equipment for business personnel based on identity authentication | |
| CN108470312B (en) | Method and device for analyzing claim case, storage medium and terminal | |
| CN112132233A (en) | Criminal personnel dangerous behavior prediction method and system based on effective influence factors | |
| CN111460312A (en) | Method and device for identifying empty-shell enterprise and computer equipment | |
| CN112419074A (en) | Vehicle insurance fraud group identification method and device | |
| CN111738843B (en) | Quantitative risk evaluation system and method using running water data | |
| CN113987190B (en) | Data quality check rule extraction method and system | |
| CN113051291A (en) | Work order information processing method, device, equipment and storage medium | |
| CN113706291A (en) | Fraud risk prediction method, device, equipment and storage medium | |
| CN115719283A (en) | Intelligent accounting management system | |
| CN114997741A (en) | Maintenance plan risk assessment method, system and equipment | |
| CN105740434A (en) | Network information scoring method and device | |
| CN113487241A (en) | Method, device, equipment and storage medium for classifying enterprise environment-friendly credit grades | |
| CN111861733B (en) | Fraud prevention and control system and method based on address fuzzy matching | |
| CN114519437B (en) | Cloud-based micro-service method and system for fault diagnosis analysis and repair reporting | |
| CN109918914A (en) | The information system attack defending ability integration assessment system and method for stratification | |
| CN114581251A (en) | Data verification method and device, computer equipment and computer readable storage medium | |
| CN114612018A (en) | Internal control risk monitoring method and system and readable storage medium | |
| CN115115322A (en) | Target group identification method, risk assessment method, apparatus, device and medium | |
| CN112307133A (en) | Security protection method and device, computer equipment and storage medium | |
| CN113723522B (en) | Abnormal user identification method and device, electronic equipment and storage medium | |
| CN113205270B (en) | Method and system for automatically generating satisfaction evaluation table and calculating evaluation score | |
| CN115482086A (en) | Rogue application behavior checking method and device and server |
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 |