Privacy Preserving Spatial Outlier Detection

Abstract

Spatial outlier detection can be applied in the finding of terrorist activities and the forecast of abnormal climate activity etc. For protecting privacy information and mining spatial outliers, we presented privacy preserving spatial outlier mining algorithm. By the definition and application of secure multiparty computation protocols based on semi-honest model, we realized the preserving of the privacy information. We utilized data mining algorithm based on privacy-preserving spatial local outlier factor (PPSLOF) to solve the mining of the spatial outlier, and used the resident linear list in memory and improved R*-tree index to decrease the communication amount, reduce the number of the input/output (I/O), and improve the retrieval velocity, so the algorithm efficiency is improved. The theory analysis shows that privacy preserving spatial outlier mining algorithm can efficiently preserve the privacy data, and efficiently mine spatial outliers.