An Integration of Prediction Model with Knowledge Base System for Motor Insurance Fraud Detection: The Case of Awash Insurance Company S.C

No Thumbnail Available



Journal Title

Journal ISSN

Volume Title


Addis Ababa University


The insurance industry has been historically a growing industry. It plays an important role in insuring the economic well being of a country. Fraud risk poses a very big challenge for the insurance sector. Business leaders are aware of the need to address this risk, but the lack of a comprehensive and integrated approach to fraud risk management continues to be a concern. Insurance fraud is very costly and has become a concern of every insurance company in recent years. Fraudulent claims account for a significant portion of all claims received by insurers, and cost billions of dollars annually. Nowadays, efforts have been made to develop models to identify potentially fraudulent claims for special investigations using data mining and knowledge base system. Data Mining is extracting information from huge set of data. In other word, data mining is inducing hidden knowledge from dataset. Knowledge based system has come across with a variety of approaches based on the knowledge representation method; Rule-based reasoning is one of the simplest and popular approach used in knowledge-based system and SWI-Prolog software is used to build a knowledge base. In this study, an attempt is made to integrate data mining with knowledge based system for fraud detection. The research has tried to apply first the clustering algorithm followed by classification techniques for developing the predictive model. K-Means clustering algorithm is employed to find the natural grouping of the different insurance claims as fraud and non-fraud and then applied PART classifier on Awash Insurance dataset. The integrator application basically created using Eclipse IDE with JDK 7 and then links the model created by PART classifier to knowledge based system. The integrator understands both the PART classifier and PROLOG and converts rules from PART to PROLOG understandable format. The experiments have been conducted following the six-step Cios et al. (2000) process model. For the experiment, the collected insurance dataset is preprocessed to remove outliers, fill in missing values and select attributes. The preprocessing phase and building the advising Knowledge base took the highest portion of the study time. Finally, the performance of the system is evaluated by preparing test cases. The cases are given to domain experts for system performance testing. For user acceptance test users evaluated the IX | P a g e system. Generally, the system has scored 81.2 % accuracy with an average Precision and Recall of 81 % and 81.2%, which is a promising result. Further researches should be done to increase the merits of integrating Data mining induced knowledge with knowledge base system. Keywords: - Insurance Fraud detection, data mining, rule based fraud detection, integration with Knowledge Base System, PART classifier



Insurance Fraud detection, data mining, rule based fraud detection, integration with Knowledge Base System, PART classifier