چکیده:
Insurance industry experts believe that fraud will be the bane of this industry. Various methods have been used over the years to detect fraud, one of which is the fuzzy expert system. Fraud detection expert systems rely on a knowledge base extracted from experts to identify fraud. However, due to the hidden nature of the phenomenon of fraud, the knowledge and judgment of experts is based on evidence and qualitative information, which often uses verbal words to describe fraudulent behavior. In the presented model, among 61 quantitative and qualitative criteria identified for detecting fraud in car body insurance, 17 criteria with high priority were categorized into eight factors based on expert opinion. The Mamdani algorithm is used for fuzzy inference in the proposed system. Finally, after designing and implementing the system in one of Iran's private insurance companies, its validity was measured through a questionnaire, and the overall validity of the system was found to be 69.45%. The calculated percentage indicates that the proposed model has a significant ability to identify fraud.
خلاصه ماشینی:
In the presented model, among the 61 quantitative and qualitative criteria identified for detecting automobile body insurance fraud, 17 criteria with high priority were categorized into eight factors based on expert opinion.
The present research aims to design an expert system to detect frauds that occur in automobile body insurance.
They first identified key factors of insurance fraud by researching and investigating experts in the insurance industry, and after assigning weights to them, they determined the most important indicators by calculating the conditional probability for each index and using regression, and proceeded to predict fraudulent claims.
They studied 72 cases, in which fraud was confirmed in 32 of them, and ultimately employed six independent variables effective in detecting fraud (Firouzi, Shokri, Kazemi & Zahedi, 1390).
Therefore, to examine the correctness or incorrectness of the claims received by the insurance company, the cases are usually categorized into two groups: legal and suspected of fraud.
Fraud Detection Criteria Extracted from Subject Literature and Expert Interviews (Refer to the page image) Prioritizing Criteria To determine the importance of each criterion for designing an expert system, the average opinion of experts regarding each criterion was obtained.
Conclusion and Suggestions This research focused on designing and presenting a fuzzy expert system with Mamdani inference method for detecting insurance fraud.
This research examined and compared the identification criteria and fraud detection methods and well justified the reasons for using the fuzzy expert system.
Fraud Detection in Car Insurance using a Fuzzy Expert System.
(
, 
, 
)
