CN-122023024-A - Insurance anti-fraud method, apparatus, equipment and storage medium

CN122023024ACN 122023024 ACN122023024 ACN 122023024ACN-122023024-A

Abstract

The embodiment of the disclosure provides insurance anti-fraud, an apparatus, equipment and a storage medium, and specifically discloses acquiring insurance associated data corresponding to an insurance request to be identified, extracting risk quantification features from the insurance associated data, matching the risk quantification features with a preset expert rule base to obtain target rules matched with the risk quantification features, wherein the expert rule base comprises preset fraud identification rules and risk scores and risk grades corresponding to the fraud identification rules, the risk grades comprise direct interception grades and indirect interception grades, and the risk quantification features are processed according to a preset membership function and a preset fuzzy rule base to obtain risk probability values, and anti-fraud treatment instructions for the insurance request are generated according to the risk scores and the risk probability values corresponding to the target rules and a preset risk threshold interval.

Inventors

Lian Shuguo
LI SITAO
LI TIANCHI
YE SIYUAN

Assignees

中国人民财产保险股份有限公司

Dates

Publication Date: 20260512
Application Date: 20260105

Claims (10)

1. A method of securing against fraud, comprising: acquiring insurance related data corresponding to an insurance request to be identified, wherein the insurance related data comprises insurance application data and insured history underwriting data; Extracting risk quantification features from the insurance related data, wherein the risk quantification features at least comprise claim settlement amount, time interval between accident risk emergence time and policy effective time and risk emergence frequency of insured persons in a preset history period; Matching the risk quantification feature with a preset expert rule base to obtain a target rule matched with the risk quantification feature, wherein the expert rule base comprises a preset fraud identification rule, and a risk score and a risk grade corresponding to the fraud identification rule; under the condition that the risk level corresponding to the target rule is an indirect interception level, processing the risk quantification feature according to a preset membership function and a preset fuzzy rule base to obtain a risk probability value; And generating an anti-fraud treatment instruction for the insurance request according to the risk score, the risk probability value and a preset risk threshold interval corresponding to the target rule, wherein the anti-fraud treatment instruction comprises automatic claim rejected, manual review or quick proposal.
2. The method according to claim 1, wherein the processing the risk quantification feature according to a preset membership function and a preset fuzzy rule base to obtain a risk probability value includes: inputting the risk quantification feature into a preset trapezoidal membership function for fuzzification treatment to obtain membership degrees of the risk quantification feature corresponding to different qualitative classifications, wherein the qualitative classifications comprise normal, suspicious and abnormal; matching the obtained membership with fuzzy rules in the fuzzy rule base, and determining a target fuzzy rule matched with the membership; and obtaining the risk probability value according to the preset weight corresponding to the target fuzzy rule and the membership corresponding to the target fuzzy rule.
3. The method according to claim 2, wherein the membership function comprises four boundary parameters, the four boundary parameters being a lower left boundary parameter, an upper right boundary parameter, and a lower right boundary parameter; The membership function is used for mapping the continuous numerical value of the risk quantification feature into membership degree changing in a preset interval through the four boundary parameters.
4. The method according to claim 2, wherein the obtaining the risk probability value according to the preset weight corresponding to the target fuzzy rule and the membership corresponding to the target fuzzy rule includes: obtaining products corresponding to the target fuzzy rules, and determining the sum of all obtained products as a first numerical value, wherein the products corresponding to the target fuzzy rules are products of membership degrees matched by the target fuzzy rules and preset weights corresponding to the target fuzzy rules; determining the sum of membership degrees matched with the target fuzzy rule as a second numerical value; and determining the ratio of the first value to the second value as the risk probability value.
5. The method according to claim 2, wherein before the risk probability value is obtained according to the preset weight corresponding to the target fuzzy rule and the membership corresponding to the target fuzzy rule, the method further comprises: The method comprises the steps of acquiring a training sample, wherein the training sample is determined based on historical insurance data and a fraud tag value corresponding to the historical insurance data, and the fraud tag value is used for indicating whether the corresponding historical insurance data is insurance fraud or non-insurance fraud; constructing a loss function by taking the difference value between the minimum risk probability value and the fraud label value as a target, and adding a regularization penalty term for weighting corresponding to the fuzzy rules in the fuzzy rule base into the loss function; And adopting a gradient descent algorithm, performing iterative optimization on the loss function by using the training sample, and determining a weight value obtained after optimization as the preset weight.
6. The method of claim 1, wherein the generating the anti-fraud treatment instruction for the insurance request according to the risk score corresponding to the target rule, the risk probability value, and a preset risk threshold interval includes: determining a first weight corresponding to the risk score and a second weight corresponding to the risk probability value according to the type of the target rule; Determining a comprehensive risk value according to the product of the risk score and the first weight and the product of the risk probability value and the second weight; and generating an anti-fraud disposal instruction for the insurance request according to the comprehensive risk value and a preset risk threshold interval.
7. A security anti-fraud device comprising: The system comprises a first acquisition module, a second acquisition module and a third acquisition module, wherein the first acquisition module is used for acquiring insurance related data corresponding to an insurance request to be identified, and the insurance related data comprises insurance application data and insured history underwriting data; the system comprises an insurance association data acquisition module, an extraction module and a risk quantification feature, wherein the insurance association data comprises insurance association data, an insurance policy management module and an insurance policy management module, wherein the insurance association data comprises insurance association data, insurance association data and insurance policy management data, and the insurance association data comprise insurance policy management data, insurance policy management data and insurance policy management data; The risk identification module is used for identifying risk quantified characteristics of the user, and acquiring a target rule matched with the risk quantified characteristics by matching the risk quantified characteristics with a preset expert rule base, wherein the expert rule base comprises a preset fraud identification rule, and a risk score and a risk grade corresponding to the fraud identification rule; the second acquisition module is used for processing the risk quantification feature according to a preset membership function and a preset fuzzy rule base under the condition that the risk level corresponding to the target rule is an indirect interception level to obtain a risk probability value; the generation module is used for generating an anti-fraud disposal instruction for the insurance request according to the risk score corresponding to the target rule, the risk probability value and a preset risk threshold interval, wherein the anti-fraud disposal instruction comprises automatic claim rejected, manual review or rapid case setting.
8. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the computer program implementing the steps of the method according to any one of claims 1 to 6 when executed by the processor.
9. A computer readable storage medium for storing computer executable instructions which when executed by a processor implement the steps of the method of any one of the preceding claims 1 to 6.
10. A computer program product, characterized in that it comprises a computer program which, when executed by a processor, implements the steps of the method according to any of the preceding claims 1 to 6.

Description

Insurance anti-fraud method, apparatus, equipment and storage medium Technical Field The present invention relates to the field of insurance technologies, and in particular, to an insurance anti-fraud method, apparatus, device, and storage medium. Background Currently, expert rule systems based on predefined logic are commonly employed to conduct insurance anti-fraud. The system can realize the rapid identification and interception of the known high-risk fraud mode by setting a series of fixed hard thresholds, such as upper limit of insurance compensation amount, frequency of risk emergence and the like. However, the expert rules generally rely on rigid judgment boundaries, so that it is difficult to effectively quantify and process scenes with gradual risk changes, and new fraud means continuously appearing in the service cannot be adaptively identified. Disclosure of Invention The invention mainly aims to provide a method, a device, equipment and a storage medium for preventing fraud, which aim to solve the problems of insufficient gradual change risk processing capacity and novel fraud recognition hysteresis caused by rigid boundaries and static rules. In a first aspect, an embodiment of the present disclosure provides a method for safeguarding against fraud, including: acquiring insurance related data corresponding to an insurance request to be identified, wherein the insurance related data comprises insurance application data and insured history underwriting data; Extracting risk quantification features from the insurance related data, wherein the risk quantification features at least comprise claim settlement amount, time interval between accident risk emergence time and policy effective time and risk emergence frequency of insured persons in a preset history period; Matching the risk quantification feature with a preset expert rule base to obtain a target rule matched with the risk quantification feature, wherein the expert rule base comprises a preset fraud identification rule, and a risk score and a risk grade corresponding to the fraud identification rule; under the condition that the risk level corresponding to the target rule is an indirect interception level, processing the risk quantification feature according to a preset membership function and a preset fuzzy rule base to obtain a risk probability value; And generating an anti-fraud treatment instruction for the insurance request according to the risk score, the risk probability value and a preset risk threshold interval corresponding to the target rule, wherein the anti-fraud treatment instruction comprises automatic claim rejected, manual review or quick proposal. In a second aspect, embodiments of the present disclosure provide a security anti-fraud device comprising: The system comprises a first acquisition module, a second acquisition module and a third acquisition module, wherein the first acquisition module is used for acquiring insurance related data corresponding to an insurance request to be identified, and the insurance related data comprises insurance application data and insured history underwriting data; the system comprises an insurance association data acquisition module, an extraction module and a risk quantification feature, wherein the insurance association data comprises insurance association data, an insurance policy management module and an insurance policy management module, wherein the insurance association data comprises insurance association data, insurance association data and insurance policy management data, and the insurance association data comprise insurance policy management data, insurance policy management data and insurance policy management data; The risk identification module is used for identifying risk quantified characteristics of the user, and acquiring a target rule matched with the risk quantified characteristics by matching the risk quantified characteristics with a preset expert rule base, wherein the expert rule base comprises a preset fraud identification rule, and a risk score and a risk grade corresponding to the fraud identification rule; the second acquisition module is used for processing the risk quantification feature according to a preset membership function and a preset fuzzy rule base under the condition that the risk level corresponding to the target rule is an indirect interception level to obtain a risk probability value; the generation module is used for generating an anti-fraud disposal instruction for the insurance request according to the risk score corresponding to the target rule, the risk probability value and a preset risk threshold interval, wherein the anti-fraud disposal instruction comprises automatic claim rejected, manual review or rapid case setting. In a third aspect, an embodiment of the present disclosure provides an electronic device comprising a processor and a memory configured to store computer-executable instructions that, when executed, cause the processor to