Detecting Fraudulent Activities in Land Record Systems: An Application of Data Mining.

摘要

Tenure security is a principle and critical factor in providing social and political stability. The major goal of this research is to assist the discovery of fraud patterns in land and property transactions using data mining techniques. The methodology starts with analysis of different fraud schemes to discover fraud patterns and their indicators in land records. A data simulator is developed to generate synthetic datasets. Different algorithms are then applied to detect fraud patterns in these datasets. Three major fraud schemes were used to validate the proposed approach (land grabbing in post conflict situations, Oklahoma Flip, and ABC-Construction).;The results have proven that data mining methods can identify fraudulent activities. However, these methods cannot be generalized to be used with all datasets. Finally, data mining can facilitate the building of fraud detection models for land transactions that can then be integrated with registration systems, and act as an alarm system.